def time_on_course(self):
"""
Estimation of time spent in course. Using self.potime,
click events are grouped by username and time sorted.
Time differences >= 10 sec and <= 3600 sec are summed.
Justification for those cutoffs can be found here:
http://cacm.acm.org/magazines/2014/4/173221-who-does-what-in-a-massive-open-online-course/fulltext
Submitted Paper: "eText Use in Blended Introductory Physics Courses: Interpreting Meaningful Interactions and the Effects of Course Structure"
Only estimates time for users with more than one event.
Parameters (generated during class initialization)
----------
None
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
fig = plt.figure(figsize=[12,6])
ax1 = fig.add_subplot(1,1,1)
### All Non-Certified
participants = self.person[self.person.certified==0].username.dropna().unique()
self.time_spent[participants].apply(np.log).hist(ax=ax1,bins=100,range=[0,16],
color='Silver',edgecolor=None,alpha=0.9,label="$Non-Certified$")
### Certified
certs = self.person[self.person.certified==1].username.dropna().unique()
self.time_spent[certs].apply(np.log).hist(ax=ax1,bins=100,range=[0,16],
color='Crimson',edgecolor=None,alpha=0.9,label="$Certified$")
ax1.set_xticks([np.log(x) for x in [1,10,60,600,3600,10*3600,100*3600]])
ax1.set_xticklabels(['1 sec','10 sec','1 min','10 min','1 hr','10 hrs','100 hrs'],rotation=40)
ax1.legend()
figsavename = self.figpath+'time_in_course_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,
xlabel='Total Time In Course',
ylabel='Count',
# title=self.nickname,
tic_size=24,
label_size=24,
gridb='y',
figsavename=figsavename+'.png')
# print figsavename
# if figsavename != None:
# dpiset = 300
# #fig.savefig('OUTPUT_Figures/%s/%s_%s.png' %(mens2_cid,figsavename,nickname), bbox_inches='tight', dpi=dpiset)
# fig.savefig('%s' % (figsavename), bbox_inches='tight', dpi=dpiset)
return None
def daily_unique_users(self):
"""
"""
fig = plt.figure(figsize=[20, 6])
ax1 = fig.add_subplot(1, 1, 1)
sgcolors = ['Silver', 'Crimson']
sglabels = ['$Non-Certified$', '$Certified$']
for sg in [0, 1]:
users = self.person[self.person.certified ==
sg].username.dropna().unique()
daily = pd.crosstab(
self.potime[self.potime.username.isin(users)].username,
self.potime.date)
daily = daily[daily > 0].count().sort_index()
daily.index = [np.datetime64(d) for d in daily.index]
daily.plot(ax=ax1,
style="-o",
ms=6,
lw=2,
color=sgcolors[sg],
rot=0,
label=sglabels[sg])
xmin = (self.cinfo['start_date'] -
np.timedelta64(2, 'W')).item().date()
xmax = (self.cinfo['end_date'] + np.timedelta64(4, 'W')).item().date()
ax1.set_xlim(xmin, xmax)
ax1 = xff.timeseries_plot_formatter(ax1, interval=1)
ax1.legend(loc=1, prop={'size': 24}, frameon=False)
ylim1 = ax1.get_ylim()[1]
ax1.vlines([
self.cinfo.start_date.item().date(),
self.cinfo.end_date.item().date()
],
0,
ylim1,
colors='Gray',
lw=1.5,
linestyles='--')
ax1.set_ylim(0, ylim1)
figsavename = self.figpath + 'daily_unique_users_' + self.nickname.replace(
'.', '_')
xff.texify(fig,
ax1,
ylabel='Unique Users',
tic_size=20,
label_size=24,
datefontsize=20,
title=self.nickname,
figsavename=figsavename + '.png')
return None
def resource_use(self, category):
"""
Number of unique resources by category.
Parameters (generated during class initialization)
----------
category: ['video','problem', etc.]
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
resources = self.caxis[self.caxis.category == category].index.unique()
usage = self.freq.filter(resources).count(axis=1)
bins = len(resources)
brange = (0, usage.max())
sgcolors = ['Silver', 'Crimson']
sglabels = ['$Non-Certified$', '$Certified$']
fig = plt.figure(figsize=[12, 6])
ax1 = fig.add_subplot(1, 1, 1)
for sg in [
0, 1
]: ### Note, later this could be all four subgroups: explored, certified, etc
subpop = self.person[self.person.certified ==
sg].username.dropna().unique()
usage[subpop].hist(
range=brange,
bins=bins,
cumulative=0, #histtype='step',
normed=True,
alpha=0.85,
edgecolor=sgcolors[sg],
color=sgcolors[sg])
# usage[subpop].hist(range=brange,bins=bins,
# cumulative=-1,histtype='step',
# normed=True,alpha=0.85,
# edgecolor=sgcolors[sg],color=sgcolors[sg],label=None)
ax1.legend(sglabels, loc=9, prop={'size': 24}, frameon=False)
figsavename = self.figpath + 'resource_use_' + category + '_dist_' + self.nickname.replace(
'.', '_')
xff.texify(fig,
ax1,
xlabel='Unique %s resources' % (category),
ylabel='Normalized Count',
title=self.nickname,
figsavename=figsavename + '.png')
return None
def enrollment_plots(self,**kwargs):
"""
Plots using the start_date from enrollment_df.
Parameters
----------
None
Output
------
Figures and respective formats.
Returns
-------
None
"""
### For JSON Data Output
jsondata = []
def date2epoch(x):
return int( (datetime.combine(x, datetime.min.time()) - datetime(1970,1,1)).total_seconds()*1000 )
for C in self.enrollment_df.columns:
fig = plt.figure(figsize=(12,6))
ax1 = fig.add_subplot(1,1,1)
self.enrollment_df[C].plot(ax=ax1,color=xff.colors['institute'],rot=0,lw=3,label=self.nickname)
ax1 = xff.timeseries_plot_formatter(ax1)
ax1.set_yticklabels([r'${0}$'.format("%d" % (y)) for y in ax1.get_yticks()])
#ax1.legend(loc=4,prop={'size':22},frameon=False,scatterpoints=1)
### Generalized Plotting functions
figsavename = self.figpath+C.replace(' ','_')+'_'+self.nickname.replace('.','_')
print figsavename
xff.texify(fig,ax1,
ylabel=C,
#title=self._xdata.course_id+' - All Registrants',
datefontsize=20,
gridb='y',
figsavename=figsavename+'.png')
### Append JSON Data
record = collections.OrderedDict()
record['key'] = C
if C == 'Enroll Count':
record['bar'] = 'true'
record['values'] = [[date2epoch(d),int(v)] for d,v in self.enrollment_df[C].iteritems()]
jsondata.append(record)
print "JSON dump currently commented out."
# str_jsondata = 'var data = '+json.dumps(jsondata)
# with open(self.figpath+'enrollment.json', 'w') as outfile:
# outfile.write(str_jsondata)
return None#self.enrollment_df
def scatter_bubble_size(self,DF,colx,coly,disc_act,figsave=False):
"""
Creates scatter plot with x=colx, y=coly.
Size of markers always proportional to disc_act (discussion activity).
Parameters (generated during class initialization)
----------
colx: column to be plotted on x-axis
coly: column to be plotted on y-axis
disc_act: column for scaling marker (bubble) size
figsave: True/False to allow exploratory analysis without saving fig.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
### Data
data = DF[[colx,coly,disc_act,'certified']].copy()
Jcolx = 0.75
Jcoly = 0.01
bmin = 1.0
bscale = 0.2
data[colx] = data[colx].apply(lambda x: x + Jcolx*(np.random.sample()-Jcolx))
data[coly] = data[coly].apply(lambda x: x + Jcoly*(np.random.sample()))
data[disc_act] = data[disc_act].fillna(1.0)
certcut = DF[DF['certified']==1].grade.min()
fig = plt.figure(figsize=[12,10])
ax1 = fig.add_subplot(1,1,1)
#Non-Certs
tmp = data[data.certified==0]
ax1.scatter(tmp[colx],tmp[coly],s=bscale*tmp[disc_act],color=xff.colors['neutral'])
#Certified
tmp = data[data.certified==1]
ax1.scatter(tmp[colx],tmp[coly],s=bscale*tmp[disc_act],color=xff.colors['institute'])
#ax1.legend(loc=5,prop={'size':18},scatterpoints=1,frameon=False)
ax1.set_xlim(-0.05,)
ax1.set_ylim(-0.05,1.05)
### Generalized Plotting functions
xff.texify(fig,ax1,xlabel=colx,ylabel=coly,
title='bubble size proportional to %s' % (disc_act.replace('_',' ')),
tic_size=20,label_size=24,datefontsize=20)
figsavename = self.figpath+'scatter_'+colx+'_'+coly+'_disc_size_'+self.nickname.replace('.','_')+'.png'
fig.savefig(figsavename, bbox_inches='tight', dpi=300)
return None
def daily_activity(self):
"""
Creates daily timeseries of discussion activity (only posts/comments/votes from forum data).
Parameters (generated during class initialization)
----------
None
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
fig = plt.figure(figsize=[20,6])
ax1 = fig.add_subplot(1,1,1)
### List of Certified Participants
certs = self.pc_plus[self.pc_plus.certified==1].user_id.unique()
certs = [str(u) for u in certs]
### Non-Certified
post_act = self.forum[ (self.forum.created_at.notnull()) & (self.forum.author_id.isin(certs)==False) ].created_at.apply(lambda x: x.date()).value_counts().sort_index()
post_act.plot(ax=ax1,style="-o",ms=6,lw=2,color=xff.colors['neutral'],label='$Non-Certified$')
#(post_act.cumsum()/10).plot(ax=ax1,style="-",ms=3,color='Orange')
### Certified
post_act = self.forum[ (self.forum.created_at.notnull()) & (self.forum.author_id.isin(certs)) ].created_at.apply(lambda x: x.date()).value_counts().sort_index()
post_act.plot(ax=ax1,style="-o",ms=6,lw=2,color=xff.colors['institute'],label='$Certified$')
xmin = (self.cinfo['start_date'] - np.timedelta64(2,'W')).item().date()
xmax = (self.cinfo['end_date'] + np.timedelta64(4,'W')).item().date()
ax1.set_xlim(xmin,xmax)
ylim1 = ax1.get_ylim()[1]
ax1.vlines([self.cinfo.start_date.item().date(),self.cinfo.end_date.item().date()], 0, ylim1, colors='Gray', lw=1.5, linestyles='--')
ax1.set_ylim(0,ylim1)
ax1 = xff.timeseries_plot_formatter(ax1,interval=1)
ax1.legend(loc=1,prop={'size':24},frameon=False)
figsavename = self.figpath+'discussion_activity_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,ylabel='Forum Text Submissions',
tic_size=20,label_size=24,
datefontsize=20,
title=self.nickname,
figsavename=figsavename+'.png')
return None
def bubble_heat_rel_week_vs_chapter(self):
"""
Bubble Heat - Unique users relative week versus chapter accessed
Color and bubble size indicate population size.
Parameters (generated during class initialization)
----------
colx: column to be plotted on x-axis
coly: column to be plotted on y-axis
disc_act: column for scaling marker (bubble) size
figsave: True/False to allow exploratory analysis without saving fig.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
### Data
data = self.potime.groupby(['rel_week','module_id']).username.apply(lambda x: len(x.unique()))
data = data.reset_index().rename(columns={0:'uniqU'})
data = pd.merge(data,self.caxis[['index','chapter']],how='left',left_on='module_id',right_index=True).dropna()
MINUSERS = 20
data = data[data.uniqU>MINUSERS]
W = data.groupby(['chapter','rel_week'])['index','uniqU'].agg([min,max]).reset_index()
fig = plt.figure(figsize=[24,16])
ax1 = fig.add_subplot(1,1,1)
c = W[('uniqU','max')]
cmhot = plt.get_cmap("CMRmap")
sc = ax1.scatter(W[('index','min')],W[('rel_week','')],s=0.95*W[('uniqU','max')],
c=c,edgecolors='None',cmap=cmhot,alpha=0.95)
cbar = plt.colorbar(sc)
cbar.set_label('$N$',rotation=90,fontsize=30)
cbar.ax.tick_params(labelsize=30)
cbar.ax.tick_params(labelsize=30)
#cbar.set_ticklabels([r'${0}$'.format("%s" % (y.get_text())) for y in cbar.ax.get_yticklabels()])
### Generalized Plotting functions
figsavename = self.figpath+'bubble_heat_rel_week_vs_chapter_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,xlabel='Course Structure Index',ylabel='Week Relative to Course Launch',
title='Unique Users visiting Chapters each Relative Week (min %d users for display)' % (MINUSERS),
tic_size=30,label_size=30,
figsavename=figsavename+'.png')
return None
def resource_use(self,category):
"""
Number of unique resources by category.
Parameters (generated during class initialization)
----------
category: ['video','problem', etc.]
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
resources = self.caxis[self.caxis.category==category].index.unique()
usage = self.freq.filter(resources).count(axis=1)
bins = len(resources)
brange = (0,usage.max())
sgcolors = ['Silver','Crimson']
sglabels = ['$Non-Certified$','$Certified$']
fig = plt.figure(figsize=[12,6])
ax1 = fig.add_subplot(1,1,1)
for sg in [0,1]: ### Note, later this could be all four subgroups: explored, certified, etc
subpop = self.person[self.person.certified==sg].username.dropna().unique()
usage[subpop].hist(range=brange,bins=bins,
cumulative=0,#histtype='step',
normed=True,alpha=0.85,
edgecolor=sgcolors[sg],color=sgcolors[sg])
# usage[subpop].hist(range=brange,bins=bins,
# cumulative=-1,histtype='step',
# normed=True,alpha=0.85,
# edgecolor=sgcolors[sg],color=sgcolors[sg],label=None)
ax1.legend(sglabels,loc=9,prop={'size':24},frameon=False)
figsavename = self.figpath+'resource_use_'+category+'_dist_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,
xlabel='Unique %s resources' % (category),
ylabel='Normalized Count',
title=self.nickname,
figsavename=figsavename+'.png')
return None
def activity_distributions(self):
"""
Creates discussion activity distributions for
users with > 0 activities (at least one action).
Parameters (generated during class initialization)
----------
None
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
gtzero = self.pc_plus[self.pc_plus.total_disc > 0]
for col in ['upvotes', 'posts', 'comments', 'total_disc']:
gtzero[col] = gtzero[col].apply(lambda x: np.log(x) if x else None)
fig = plt.figure(figsize=[12, 7])
ax1 = fig.add_subplot(1, 1, 1)
### Non-Certs
gtzero[gtzero.certified == 0][col].hist(
bins=50,
range=[0, 9],
log=True,
color=xff.colors['neutral'],
edgecolor=xff.colors['neutral'])
### Certs
gtzero[gtzero.certified == 1][col].hist(
bins=50,
range=[0, 9],
log=True,
color=xff.colors['institute'],
edgecolor=xff.colors['institute'],
alpha=0.8)
tics = [1, 10, 100, 1000, 10000]
ax1.set_xticks([np.log(x) for x in tics])
ax1.set_xticklabels(tics)
ax1.set_xlim([0, 8])
xff.texify(fig, ax1, xlabel=col.replace('_', ' '), ylabel='Count')
figsavename = self.figpath + 'disc_act_distribution_' + col + '_' + self.nickname + '.png'
fig.savefig(figsavename, bbox_inches='tight', dpi=300)
return None
def activity_distributions(self, columns=[], **kwargs):
"""
Activity Distributions from person_course collection using Pandas built in hist function.
There are several columns which can be plotted as distributions:
features = ['ndays_act', 'nevents', 'nforum_events', 'nforum_posts', 'nplay_video', 'nproblem_check', 'nprogcheck']
Course Reports focus on nevents, ndays_act, and nchapters
Parameters
----------
columns : list of column names
Typical column names - ['ndays_act', 'nevents', 'nforum_events',
'nforum_posts', 'nplay_video', 'nproblem_check',
'nprogcheck']
Returns
-------
None
"""
data = self.person.filter(items=columns)
for col in columns:
if col not in self.person:
columns.remove(col)
print "Some of the specified columns do not exist in the person_course collection."
print col
if len(columns) == 0:
print "No specified columns available."
return None
for column in data.columns:
### Plot
fig = plt.figure(figsize=(12, 7))
ax1 = fig.add_subplot(1, 1, 1)
check = self.distribution_logic(ax1, column, True)
### Generalized Plotting functions
figsavename = self.figpath + 'distribution_' + column.replace(
' ', '_') + '_' + self.nickname.replace('.', '_')
print figsavename
xff.texify(fig, ax1, figsavename=figsavename + '.png')
return None
def activity_distributions(self,columns=[],**kwargs):
"""
Activity Distributions from person_course collection using Pandas built in hist function.
There are several columns which can be plotted as distributions:
features = ['ndays_act', 'nevents', 'nforum_events', 'nforum_posts', 'nplay_video', 'nproblem_check', 'nprogcheck']
Course Reports focus on nevents, ndays_act, and nchapters
Parameters
----------
columns : list of column names
Typical column names - ['ndays_act', 'nevents', 'nforum_events',
'nforum_posts', 'nplay_video', 'nproblem_check',
'nprogcheck']
Returns
-------
None
"""
data = self.person.filter(items=columns)
for col in columns:
if col not in self.person:
columns.remove(col)
print "Some of the specified columns do not exist in the person_course collection."
print col
if len(columns) == 0:
print "No specified columns available."
return None
for column in data.columns:
### Plot
fig = plt.figure(figsize=(12,7))
ax1 = fig.add_subplot(1,1,1)
check = self.distribution_logic(ax1,column,True)
### Generalized Plotting functions
figsavename = self.figpath+'distribution_'+column.replace(' ','_')+'_'+self.nickname.replace('.','_')
print figsavename
xff.texify(fig,ax1,figsavename=figsavename+'.png')
return None
def activity_distributions(self):
"""
Creates discussion activity distributions for
users with > 0 activities (at least one action).
Parameters (generated during class initialization)
----------
None
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
gtzero = self.pc_plus[self.pc_plus.total_disc>0]
for col in ['upvotes','posts','comments','total_disc']:
gtzero[col] = gtzero[col].apply(lambda x: np.log(x) if x else None)
fig = plt.figure(figsize=[12,7])
ax1 = fig.add_subplot(1,1,1)
### Non-Certs
gtzero[gtzero.certified==0][col].hist(bins=50,range=[0,9],log=True,
color=xff.colors['neutral'],
edgecolor=xff.colors['neutral'])
### Certs
gtzero[gtzero.certified==1][col].hist(bins=50,range=[0,9],log=True,
color=xff.colors['institute'],
edgecolor=xff.colors['institute'],
alpha=0.8)
tics = [1,10,100,1000,10000]
ax1.set_xticks([np.log(x) for x in tics])
ax1.set_xticklabels(tics)
ax1.set_xlim([0,8])
xff.texify(fig,ax1,xlabel=col.replace('_',' '),ylabel='Count')
figsavename = self.figpath+'disc_act_distribution_'+col+'_'+self.nickname+'.png'
fig.savefig(figsavename, bbox_inches='tight', dpi=300)
return None
def daily_unique_users(self):
"""
"""
fig = plt.figure(figsize=[20,6])
ax1 = fig.add_subplot(1,1,1)
sgcolors = ['Silver','Crimson']
sglabels = ['$Non-Certified$','$Certified$']
for sg in [0,1]:
users = self.person[self.person.certified==sg].username.dropna().unique()
daily = pd.crosstab(self.potime[self.potime.username.isin(users)].username,self.potime.date)
daily = daily[daily>0].count().sort_index()
daily.index = [np.datetime64(d) for d in daily.index]
daily.plot(ax=ax1,style="-o",ms=6,lw=2,color=sgcolors[sg],rot=0,label=sglabels[sg])
xmin = (self.cinfo['start_date'] - np.timedelta64(2,'W')).item().date()
xmax = (self.cinfo['end_date'] + np.timedelta64(4,'W')).item().date()
ax1.set_xlim(xmin,xmax)
ax1 = xff.timeseries_plot_formatter(ax1, interval=1)
ax1.legend(loc=1,prop={'size':24},frameon=False)
ylim1 = ax1.get_ylim()[1]
ax1.vlines([self.cinfo.start_date.item().date(),self.cinfo.end_date.item().date()], 0, ylim1, colors='Gray', lw=1.5, linestyles='--')
ax1.set_ylim(0,ylim1)
figsavename = self.figpath+'daily_unique_users_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,ylabel='Unique Users',
tic_size=20,label_size=24,
datefontsize=20,
title=self.nickname,
figsavename=figsavename+'.png')
return None
def grade_distribution(self, **kwargs):
'''
Simple grade distribution.
Parameters
----------
None
Output
------
Figures and respective formats.
Returns
-------
None
'''
NVD3 = kwargs.get('NVD3', False)
if self.person.grade.count() < 50:
print "Not enough grade information. Return without grade distribution."
return None
bins = 50
hmin = 0.0 #self.person['grade'].min()
hmax = 1.0 #self.person['grade'].max()
glowbound = 0.1
### Plot
fig = plt.figure(figsize=(12, 7))
ax1 = fig.add_subplot(1, 1, 1)
self.person[(self.person.grade >= glowbound)
& (self.person.certified == 0)]['grade'].hist(
ax=ax1,
bins=bins,
range=(hmin, hmax),
log=False,
cumulative=0,
color=xff.colors['neutral'],
edgecolor=xff.colors['neutral'])
self.person[(self.person.grade >= glowbound)
& (self.person.certified == 1)]['grade'].hist(
ax=ax1,
bins=bins,
range=(hmin, hmax),
log=False,
cumulative=0,
color=xff.colors['institute'],
edgecolor=xff.colors['institute'],
alpha=0.8)
xlab = 'Grade'
#ax1.set_xlabel(r'%s' % (xlab),fontdict={'fontsize': 30,'style': 'oblique'})
#ax1.set_ylabel(r'Count (log scale)',fontdict={'fontsize': 30,'style': 'oblique'})
ax1.set_xlim(0, hmax)
ax1.set_ylim(1, )
ax1.legend(['$Non-Certified$', '$Certified$'],
loc=1,
prop={'size': 24},
frameon=False)
ax1.set_xticklabels([r'$%.1f$' % x for x in ax1.get_xticks()],
fontsize=30)
ax1.set_yticklabels([r'$%d$' % y for y in ax1.get_yticks()],
fontsize=30)
### Generalized Plotting functions
figsavename = self.figpath + 'grade_distribution_' + self.nickname.replace(
'.', '_')
xff.texify(fig,
ax1,
xlabel='Grade (> %.2f)' % (glowbound),
ylabel='Count',
gridb='y',
figsavename=figsavename + '.png')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### Data
bins = 50
hmin = 0.0
hmax = 1.0
DATA = self.person[(self.person.grade >= glowbound)]
Y1, X1 = np.histogram(DATA[DATA.certified == 0].grade.values,
bins=bins,
range=(hmin, hmax))
Y2, X2 = np.histogram(DATA[DATA.certified == 1].grade.values,
bins=bins,
range=(hmin, hmax))
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath + 'interactive_grade_distribution_' + self.nickname + '.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Grade Distribution: %s" % self._xdata.course_id
chart = multiBarChart(name=title, height=400)
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(X1)
extra_serie = {
"tooltip": {
"y_start": "",
"y_end": ""
},
"color": xff.colors['neutral']
}
chart.add_serie(name="Count", y=Y1, x=X1, extra=extra_serie)
extra_serie = {
"tooltip": {
"y_start": "",
"y_end": ""
},
"color": xff.colors['institute']
}
chart.add_serie(name="Duration", y=Y2, x=X2, extra=extra_serie)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def country_of_origin(self, **kwargs):
"""
Creates figures for the top "ccnum" of enrolled countries.
Parameters (generated during class initialization)
----------
ccnum = number of requested countries to be plotted. Max 25 for plotting issues.
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3', False)
ccnum = kwargs.get('ccnum', 10)
cc = self.person.final_cc.value_counts().order(ascending=False)
if self.person[self.person.certified ==
1].username.count() > self.mincerts:
certs = self.person[self.person.certified ==
1].final_cc.value_counts()
else:
certs = pd.Series(index=cc.index)
cc = pd.concat([cc, certs],
join='inner',
axis=1,
keys=['$Non-Certified$', '$Certified$'])
cc = cc.sort('$Non-Certified$', ascending=False)[0:ccnum]
perc = 100. * cc / cc.sum()
perc = perc.apply(lambda x: np.round(x, 1))
#print perc
fig = plt.figure(figsize=(12, 6))
ax1 = fig.add_subplot(1, 1, 1)
perc.plot(
ax=ax1,
kind='bar',
color=[xff.colors['neutral'], xff.colors['institute']],
rot=40,
)
### Plot Details
ax1.set_xticklabels([r'$%s$' % x for x in perc.index])
ax1.set_yticklabels(
[r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],
fontsize=30)
ax1.legend(loc=1, prop={'size': 28}, frameon=False)
### Generalized Plotting functions
figsavename = self.figpath + 'country_geoloc_distribution_' + self.nickname.replace(
'.', '_')
print figsavename
xff.texify(fig,
ax1,
xlabel='Country Code',
ylabel=None,
figsavename=figsavename + '.png')
### Output JSON Records
#cc.name = 'value'
#cc = cc.reset_index().rename(columns={'index':'label'})
#cc.dropna().to_json(figsavename+'.json',orient='records')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath + 'interactive_country_distribution_' + self.nickname + '.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Education Level Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype,
height=350,
x_axis_format="",
y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(perc)
X = perc.index #list(range(nb_element))
Y1 = perc.ix[:, '$Non-Certified$'].values
Y2 = perc.ix[:, '$Certified$'].values
### Series 1
extra_serie1 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['neutral'],
"format": ".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['institute'],
"format": ".1f"
}
chart.add_serie(name="Certificate Earners",
y=Y2,
x=X,
extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def gender(self,**kwargs):
"""
Creates gender distribution figures.
Parameters (generated during class initialization)
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3',False)
### Removes those users not having the option to fill in edX registration data.
trim_data = self.person[(self.person.registered==1) & (self.person.user_id>156633)]
### Data
gdict = {'f': "$Female$",'m': "$Male$",'o':"$Other$"}
glist = ['$Female$','$Male$']
### Munge and Plot
gender = trim_data.gender.dropna().apply(lambda x: gdict[x]).value_counts()
#print gender
certs = trim_data[trim_data.certified==1]
if certs.username.count() > self.mincerts:
certs = certs.gender.dropna().apply(lambda x: gdict[x]).value_counts()
else:
certs = pd.Series(index=gender.index)
gender = pd.concat([gender,certs],join='inner',axis=1,keys=['$Non-Certified$','$Certified$'])
gender = 100.*gender/gender.sum()
gender = gender.apply(lambda x: np.round(x,1))
fig = plt.figure(figsize=(12,6))
ax1 = fig.add_subplot(1,1,1)
gender.ix[glist,:].plot(ax=ax1,kind='bar',color=[xff.colors['neutral'],xff.colors['institute']],rot=0)
### Plot Details
ax1.set_xticklabels([r'%s' % x for x in glist])
ax1.set_yticklabels([r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],fontsize=30)
ax1.legend(loc=2,prop={'size':28},frameon=False)
### Generalized Plotting functions
figsavename = self.figpath+'gender_distribution_'+self.nickname.replace('.','_')
print figsavename
xff.texify(fig,ax1,xlabel=None,ylabel='Count',figsavename=figsavename+'.png')
# ### Output JSON Records
# gender.name = 'value'
# gender = gender.reset_index().rename(columns={'index':'label'})
# gender.dropna().to_json(figsavename+'.json',orient='records')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
'http://nvd3.org/examples/pie.html'
X = [ x.replace('$','') for x in gender.index ]
Y1 = gender.ix[glist,'$Non-Certified$'].values
Y2 = gender.ix[glist,'$Certified$'].values
#----------------------------------------------------------------
### BAR Chart
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath+'interactive_gender_distribution_'+self.nickname+'.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Gender Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype, height=350, x_axis_format="", y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(gender.ix[glist,:])
### Series 1
extra_serie1 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['neutral'],
"format":".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['institute'],
"format":".1f"
}
chart.add_serie(name="Certificate Earners", y=Y2, x=X, extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
#----------------------------------------------------------------
### Pie Chart
from nvd3 import pieChart
### Output File
figsavename = self.figpath+'interactive_gender_piechart_'+self.nickname+'.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Gender Pie Chart: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = pieChart(name=charttype, color_category='category20c', height=400, width=400)
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
extra_serie = {"tooltip": {"y_start": "", "y_end": " certified"}}
chart.add_serie(y=Y1, x=X, extra=extra_serie)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def bubble_heat_rel_week_vs_chapter(self):
"""
Bubble Heat - Unique users relative week versus chapter accessed
Color and bubble size indicate population size.
Parameters (generated during class initialization)
----------
colx: column to be plotted on x-axis
coly: column to be plotted on y-axis
disc_act: column for scaling marker (bubble) size
figsave: True/False to allow exploratory analysis without saving fig.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
### Data
data = self.potime.groupby(
['rel_week',
'module_id']).username.apply(lambda x: len(x.unique()))
data = data.reset_index().rename(columns={0: 'uniqU'})
data = pd.merge(data,
self.caxis[['index', 'chapter']],
how='left',
left_on='module_id',
right_index=True).dropna()
MINUSERS = 20
data = data[data.uniqU > MINUSERS]
W = data.groupby(['chapter',
'rel_week'])['index',
'uniqU'].agg([min, max]).reset_index()
fig = plt.figure(figsize=[24, 16])
ax1 = fig.add_subplot(1, 1, 1)
c = W[('uniqU', 'max')]
cmhot = plt.get_cmap("CMRmap")
sc = ax1.scatter(W[('index', 'min')],
W[('rel_week', '')],
s=0.95 * W[('uniqU', 'max')],
c=c,
edgecolors='None',
cmap=cmhot,
alpha=0.95)
cbar = plt.colorbar(sc)
cbar.set_label('$N$', rotation=90, fontsize=30)
cbar.ax.tick_params(labelsize=30)
cbar.ax.tick_params(labelsize=30)
#cbar.set_ticklabels([r'${0}$'.format("%s" % (y.get_text())) for y in cbar.ax.get_yticklabels()])
### Generalized Plotting functions
figsavename = self.figpath + 'bubble_heat_rel_week_vs_chapter_' + self.nickname.replace(
'.', '_')
xff.texify(
fig,
ax1,
xlabel='Course Structure Index',
ylabel='Week Relative to Course Launch',
title=
'Unique Users visiting Chapters each Relative Week (min %d users for display)'
% (MINUSERS),
tic_size=30,
label_size=30,
figsavename=figsavename + '.png')
return None
def time_on_course(self):
"""
Estimation of time spent in course. Using self.potime,
click events are grouped by username and time sorted.
Time differences >= 10 sec and <= 3600 sec are summed.
Justification for those cutoffs can be found here:
http://cacm.acm.org/magazines/2014/4/173221-who-does-what-in-a-massive-open-online-course/fulltext
Submitted Paper: "eText Use in Blended Introductory Physics Courses: Interpreting Meaningful Interactions and the Effects of Course Structure"
Only estimates time for users with more than one event.
Parameters (generated during class initialization)
----------
None
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
fig = plt.figure(figsize=[12, 6])
ax1 = fig.add_subplot(1, 1, 1)
### All Non-Certified
participants = self.person[self.person.certified ==
0].username.dropna().unique()
self.time_spent[participants].apply(np.log).hist(
ax=ax1,
bins=100,
range=[0, 16],
color='Silver',
edgecolor=None,
alpha=0.9,
label="$Non-Certified$")
### Certified
certs = self.person[self.person.certified ==
1].username.dropna().unique()
self.time_spent[certs].apply(np.log).hist(ax=ax1,
bins=100,
range=[0, 16],
color='Crimson',
edgecolor=None,
alpha=0.9,
label="$Certified$")
ax1.set_xticks(
[np.log(x) for x in [1, 10, 60, 600, 3600, 10 * 3600, 100 * 3600]])
ax1.set_xticklabels([
'1 sec', '10 sec', '1 min', '10 min', '1 hr', '10 hrs', '100 hrs'
],
rotation=40)
ax1.legend()
figsavename = self.figpath + 'time_in_course_' + self.nickname.replace(
'.', '_')
xff.texify(
fig,
ax1,
xlabel='Total Time In Course',
ylabel='Count',
# title=self.nickname,
tic_size=24,
label_size=24,
gridb='y',
figsavename=figsavename + '.png')
# print figsavename
# if figsavename != None:
# dpiset = 300
# #fig.savefig('OUTPUT_Figures/%s/%s_%s.png' %(mens2_cid,figsavename,nickname), bbox_inches='tight', dpi=dpiset)
# fig.savefig('%s' % (figsavename), bbox_inches='tight', dpi=dpiset)
return None
def grade_distribution(self,**kwargs):
'''
Simple grade distribution.
Parameters
----------
None
Output
------
Figures and respective formats.
Returns
-------
None
'''
NVD3 = kwargs.get('NVD3',False)
if self.person.grade.count() < 50:
print "Not enough grade information. Return without grade distribution."
return None
bins = 50
hmin = 0.0 #self.person['grade'].min()
hmax = 1.0 #self.person['grade'].max()
glowbound = 0.1
### Plot
fig = plt.figure(figsize=(12,7))
ax1 = fig.add_subplot(1,1,1)
self.person[(self.person.grade>=glowbound) & (self.person.certified==0)]['grade'].hist(ax=ax1,bins=bins,range=(hmin,hmax),log=False,cumulative=0,
color=xff.colors['neutral'],edgecolor=xff.colors['neutral'])
self.person[(self.person.grade>=glowbound) & (self.person.certified==1)]['grade'].hist(ax=ax1,bins=bins,range=(hmin,hmax),log=False,cumulative=0,
color=xff.colors['institute'],edgecolor=xff.colors['institute'],alpha=0.8)
xlab = 'Grade'
#ax1.set_xlabel(r'%s' % (xlab),fontdict={'fontsize': 30,'style': 'oblique'})
#ax1.set_ylabel(r'Count (log scale)',fontdict={'fontsize': 30,'style': 'oblique'})
ax1.set_xlim(0,hmax)
ax1.set_ylim(1,)
ax1.legend(['$Non-Certified$','$Certified$'],loc=1,prop={'size':24},frameon=False)
ax1.set_xticklabels([r'$%.1f$' % x for x in ax1.get_xticks()],fontsize=30)
ax1.set_yticklabels([r'$%d$' % y for y in ax1.get_yticks()],fontsize=30)
### Generalized Plotting functions
figsavename = self.figpath+'grade_distribution_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,
xlabel='Grade (> %.2f)' % (glowbound),
ylabel='Count',
gridb='y',
figsavename=figsavename+'.png')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### Data
bins = 50
hmin = 0.0
hmax = 1.0
DATA = self.person[(self.person.grade>=glowbound)]
Y1,X1 = np.histogram(DATA[DATA.certified==0].grade.values,bins=bins,range=(hmin,hmax))
Y2,X2 = np.histogram(DATA[DATA.certified==1].grade.values,bins=bins,range=(hmin,hmax))
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath+'interactive_grade_distribution_'+self.nickname+'.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Grade Distribution: %s" % self._xdata.course_id
chart = multiBarChart(name=title, height=400)
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(X1)
extra_serie = {"tooltip": {"y_start": "", "y_end": ""},
"color":xff.colors['neutral']
}
chart.add_serie(name="Count", y=Y1, x=X1, extra=extra_serie)
extra_serie = {"tooltip": {"y_start": "", "y_end": ""},
"color":xff.colors['institute']
}
chart.add_serie(name="Duration", y=Y2, x=X2, extra=extra_serie)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def last_day_plots(self, **kwargs):
"""
Plots using the last_event date from enrollment_df.
Parameters
----------
None
Output
------
Figures and respective formats.
Returns
-------
None
"""
col = self.lastact_df.columns[0]
tmp = self.lastact_df[col][self.lastact_df[col] > 100]
mindate = tmp.index[0] - timedelta(days=21)
maxdate = tmp.index[-1] + timedelta(days=21)
#print mindate,maxdate
### For JSON Data Output
jsondata = []
def date2epoch(x):
return int((datetime.combine(x, datetime.min.time()) -
datetime(1970, 1, 1)).total_seconds() * 1000)
for C in self.lastact_df.columns:
#print C,self.lastact_df[C].idxmin(),self.lastact_df[C].idxmax()
fig = plt.figure(figsize=(12, 6))
ax1 = fig.add_subplot(1, 1, 1)
self.lastact_df[C].plot(ax=ax1,
color=xff.colors['institute'],
rot=0,
lw=3,
label=self.nickname)
ax1.set_xlim(mindate, maxdate)
ax1 = xff.timeseries_plot_formatter(ax1)
ax1.set_yticklabels(
[r'${0}$'.format("%d" % (y)) for y in ax1.get_yticks()])
#ax1.legend(loc=4,prop={'size':22},frameon=False,scatterpoints=1)
### Generalized Plotting functions
figsavename = self.figpath + C.replace(
' ', '_') + '_' + self.nickname.replace('.', '_')
print figsavename
xff.texify(
fig,
ax1,
ylabel=C,
#title=self._xdata.course_id+' - All Registrants',
datefontsize=20,
gridb='y',
figsavename=figsavename + '.png')
### Append JSON Data
record = collections.OrderedDict()
record['key'] = C
if C == 'Last Activity Count':
record['bar'] = 'true'
record['values'] = [[date2epoch(d), int(v)]
for d, v in self.lastact_df[C].iteritems()]
jsondata.append(record)
print "JSON dump currently commented out."
# str_jsondata = 'var data = '+json.dumps(jsondata)
# with open(self.figpath+'lastday.json', 'w') as outfile:
# outfile.write(str_jsondata)
return None
def andrew_ho_diagram(self, **kwargs):
"""
Plot showing the intersection of enrollment populations by categories
defined in the 2013 (published 2014) course reports.
http://odl.mit.edu/mitx-working-papers/
Parameters
----------
None
Output
------
Figures and respective formats.
Returns
-------
None
"""
### Registration Types
self.person['Only Registered'] = 0
self.person['Only Viewed'] = 0
self.person['Only Explored'] = 0
g = self.person ### This is a bit silly, but keeps lines short despite the conditions.
### Create disjoint groups (note g and self.person are the same)
#Only Registered: A - (B+C+D)
reg_list = g[(g['registered'] == 1) & (g['viewed'] == 0) &
(g['explored'] == 0) & (g['certified'] == 0)].user_id
self.person.ix[self.person[self.person.user_id.isin(reg_list)].index,
'Only Registered'] = 1
#Only Viewed: B - (C+D)
view_list = g[(g.user_id.isin(reg_list) == False) & (g['viewed'] == 1)
& (g['explored'] == 0) & (g['certified'] == 0)].user_id
self.person.ix[self.person[self.person.user_id.isin(view_list)].index,
'Only Viewed'] = 1
#Only Explored: C - (D)
exp_list = g[(g.user_id.isin(reg_list) == False)
& (g.user_id.isin(view_list) == False) &
(g['explored'] == 1) & (g['certified'] == 0)].user_id
self.person.ix[self.person[self.person.user_id.isin(exp_list)].index,
'Only Explored'] = 1
### Figure
fig = plt.figure(figsize=(12, 8))
ax1 = fig.add_subplot(111)
#Circles
SCALE = 1000
expratio = SCALE * 100. * len(
self.person[self.person['Only Explored'] == 1]) / len(
self.person[self.person['registered'] == 1])
#print expratio
if expratio > 17000:
expratio = 1000
certratio = SCALE * 100. * len(
self.person[self.person['certified'] == 1]) / len(
self.person[self.person['registered'] == 1])
#print certratio
if certratio < 1000:
certratio = 1000
# csize = 1000 # Ratio and csize give relative size of explored and certified circles.
# # expl = len(self.person[self.person['Only Explored']==1])
# # cert = len(self.person[self.person['certified']==1])
ax1.scatter([0.45], [0.5],
s=expratio,
edgecolor='Black',
lw=2,
color='white',
alpha=0.8)
ax1.scatter([0.50], [0.5],
s=certratio,
edgecolor=xff.colors['institute'],
lw=2,
color='white',
alpha=0.8)
#Rectangles
rect1 = matplotlib.patches.Rectangle((0, 0),
1,
1,
fill=False,
fc='white',
ec='black',
lw=2)
rect2 = matplotlib.patches.Rectangle((0.15, 0.15),
0.7,
0.7,
fill=False,
fc='white',
ec='black',
lw=2)
ax1.add_patch(rect1)
ax1.add_patch(rect2)
FS = 25
### Only registerd
x = len(self.person[self.person['Only Registered'] == 1])
y = 100.0 * x / len(self.person[self.person['registered'] == 1])
ax1.text(0.05,
0.925,
'$Only\ registered:\ %d\ \ (%.1f\%%)$' % (x, y),
fontsize=FS,
ha='left')
x = len(self.person[self.person['Only Viewed'] == 1])
y = 100.0 * x / len(self.person[self.person['registered'] == 1])
ax1.text(0.2,
0.775,
'$Only\ viewed:\ %d\ \ (%.1f\%%)$' % (x, y),
fontsize=FS,
ha='left')
x = len(self.person[self.person['Only Explored'] == 1])
y = 100.0 * x / len(self.person[self.person['registered'] == 1])
ax1.text(0.175,
0.2,
'$Only\ explored:$\n$%d\ \ (%.1f\%%)$' % (x, y),
fontsize=FS,
ha='left')
x = len(self.person[self.person['certified'] == 1])
y = 100.0 * x / len(self.person[self.person['registered'] == 1])
ax1.text(0.625,
0.35,
'$Certified:$\n$%d\ \ (%.1f\%%)$' % (x, y),
fontsize=FS,
ha='left')
ax1.set_xlim([-0.01, 1.01])
ax1.set_ylim([-0.01, 1.01])
## suppress spines
for key in ax1.spines.keys():
ax1.spines[key].set_color('none')
ax1.axes.get_xaxis().set_visible(False)
ax1.axes.get_yaxis().set_visible(False)
### Generalized Plotting functions
figsavename = self.figpath + 'AHO_Diagram' + '_' + self.nickname.replace(
'.', '_')
xff.texify(fig, ax1, figsavename=figsavename + '.png')
# ### Package data
# jsonout = pd.Series()
# for sg in ['Only Registered','Only Viewed','Only Explored','certified']:
# count = len(self.person[self.person[sg]==1].username.unique())
# jsonout.set_value(sg.capitalize()+'(N=%d)'%(count),count)
# jsonout = 100.0*jsonout/jsonout.sum()
# jsondata = []
# for l,v in jsonout.iteritems():
# record = collections.OrderedDict()
# record['label'] = l
# record['value'] = v
# jsondata.append(record)
# print "JSON dump currently commented out."
# # str_jsondata = 'var data = '+json.dumps(jsondata)
# # with open(self.figpath+'percent_participanttypes.json', 'w') as outfile:
# # outfile.write(str_jsondata)
return None
def content_touches_viz(self,horiz_w_data,certified):
'''
'''
fig = plt.figure(figsize=(32,12))
fig.subplots_adjust(hspace=.1)
#plt.rcParams.update({'font.size': 20})
ax1 = fig.add_subplot(2,1,1)
ax2 = fig.add_subplot(2,1,2)
#ax1.plot(tmp['order'],tmp['uniqU'],'o')
bars1 = ax1.bar(horiz_w_data['order'],horiz_w_data['users'],3.0,alpha=0.8,edgecolor='none')
#Colors the bars (this is where reindexing matters)
for i,b in enumerate(bars1):
flipi = i#len(tmp.index)-1-i
if horiz_w_data.color[flipi] != 'Pink':
bars1[i].set_facecolor(horiz_w_data.color[flipi])
bars1[i].set_edgecolor(horiz_w_data.color[flipi])
else:
bars1[i].set_facecolor('none')
bars1[i].set_edgecolor('none')
#ax1.plot(tmp['order'],tmp.uniqU,'-o',color='Silver',alpha=0.8)
ax1.set_xlabel('Course Structure Index')
ax1.set_ylabel('Unique Users')
#ax1.set_xlim(0,2100)
bars2 = ax2.bar(horiz_w_data['order'],horiz_w_data.scale,2.5,edgecolor='none')
#Colors the bars (this is where reindexing matters)
for i,b in enumerate(bars2):
flipi = i#len(vert.index)-1-i
bars2[i].set_facecolor(horiz_w_data.color[flipi])
invert = True # Choose whether to have the CC plot left or right oriented
ha = 'left'
if invert == True:
ax2.invert_yaxis()
ha = 'right'
ax2.axes.get_xaxis().set_ticks([])
#ax2.set_xlim(0,500)
ax2.axes.get_yaxis().set_ticks([])
#ax2.set_xlim(0,2100)
ax2.set_ylim(7,-1)
#fig.patch.set_visible(False)
ax2.axis('off')
#!!!!! x limits must be set together
ax1.set_xlim(ax2.get_xlim()[0],ax2.get_xlim()[1])
#ax1.set_ylim(0,7500)
xff.texify(fig,ax1,tic_size=32,label_size=32)
dpiset = 300
if certified==True:
figsavename = self.figpath+'content_touches_horizontal_certified_'+self.nickname.replace('.','_')+'.png'
else:
figsavename = self.figpath+'content_touches_horizontal_'+self.nickname.replace('.','_')+'.png'
fig.savefig(figsavename, bbox_inches='tight', dpi=dpiset)
return None
def country_of_origin(self,**kwargs):
"""
Creates figures for the top "ccnum" of enrolled countries.
Parameters (generated during class initialization)
----------
ccnum = number of requested countries to be plotted. Max 25 for plotting issues.
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3',False)
ccnum = kwargs.get('ccnum',10)
cc = self.person.final_cc.value_counts().order(ascending=False)
if self.person[self.person.certified==1].username.count() > self.mincerts:
certs = self.person[self.person.certified==1].final_cc.value_counts()
else:
certs = pd.Series(index=cc.index)
cc = pd.concat([cc,certs],join='inner',axis=1,keys=['$Non-Certified$','$Certified$'])
cc = cc.sort('$Non-Certified$',ascending=False)[0:ccnum]
perc = 100.*cc/cc.sum()
perc = perc.apply(lambda x: np.round(x,1))
#print perc
fig = plt.figure(figsize=(12,6))
ax1 = fig.add_subplot(1,1,1)
perc.plot(ax=ax1,kind='bar',color=[xff.colors['neutral'],xff.colors['institute']],rot=40,)
### Plot Details
ax1.set_xticklabels([r'$%s$' % x for x in perc.index])
ax1.set_yticklabels([r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],fontsize=30)
ax1.legend(loc=1,prop={'size':28},frameon=False)
### Generalized Plotting functions
figsavename = self.figpath+'country_geoloc_distribution_'+self.nickname.replace('.','_')
print figsavename
xff.texify(fig,ax1,xlabel='Country Code',ylabel=None,figsavename=figsavename+'.png')
### Output JSON Records
#cc.name = 'value'
#cc = cc.reset_index().rename(columns={'index':'label'})
#cc.dropna().to_json(figsavename+'.json',orient='records')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath+'interactive_country_distribution_'+self.nickname+'.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Education Level Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype, height=350, x_axis_format="", y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(perc)
X = perc.index #list(range(nb_element))
Y1 = perc.ix[:,'$Non-Certified$'].values
Y2 = perc.ix[:,'$Certified$'].values
### Series 1
extra_serie1 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['neutral'],
"format":".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['institute'],
"format":".1f"
}
chart.add_serie(name="Certificate Earners", y=Y2, x=X, extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def level_of_education(self,**kwargs):
'''
Plot Level of Education Attained; typically taken from the edX enrollment questionairre.
'''
"""
Creates distribution of highest level of education attained.
Parameters (generated during class initialization)
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3',False)
### Level of Education (LoE)
### Data
eddict = {'el': "Less\ than$\n$ Secondary",'jhs': "Less\ than$\n$ Secondary",'none':"Less\ than$\n$ Secondary",
'hs':"Secondary",'a':"Secondary",
'b':"Bachelor\'s",
'm': "Master\'s",
'p_se': "Doctorate",'p_oth': "Doctorate",'p': "Doctorate",
'other': None,'NA':None,'nan':None,
}
edlist = ["Less\ than$\n$ Secondary","Secondary","Bachelor\'s","Master\'s","Doctorate"]
trim_data = self.person[(self.person.registered==1) & (self.person.user_id>156633)]
edlevels = trim_data.LoE.apply(lambda x: eddict[str(x)] if x in eddict.keys() else None).value_counts()[edlist]
if trim_data[trim_data.certified==1].username.count() > self.mincerts:
certs = trim_data[trim_data.certified==1].LoE.apply(lambda x: eddict[str(x)] if x in eddict.keys() else None).value_counts()[edlist]
else:
certs = pd.Series(index=edlevels.index)
edlevels = pd.concat([edlevels,certs],join='inner',axis=1,keys=['$Non-Certified$','$Certified$'])
edlevels = 100.*edlevels/edlevels.sum()
edlevels = edlevels.apply(lambda x: np.round(x,1))
#print edlevels
#Plot
fig = plt.figure(figsize=(12,6))
ax1 = fig.add_subplot(1,1,1)
edlevels.plot(ax=ax1,kind='bar',color=[xff.colors['neutral'],xff.colors['institute']],rot=40)
### Plot Details
ax1.set_xticklabels([r'$%s$' % x for x in edlist])
ax1.set_yticklabels([r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],fontsize=30)
ax1.legend(loc=2,prop={'size':22},frameon=False)
### Generalized Plotting functions
figsavename = self.figpath+'loe_distribution_'+self.nickname.replace('.','_')
print figsavename
xff.texify(fig,ax1,xlabel=None,ylabel=None,figsavename=figsavename+'.png')
### Output JSON Records
#cc.name = 'value'
#cc = cc.reset_index().rename(columns={'index':'label'})
#cc.dropna().to_json(figsavename+'.json',orient='records')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath+'interactive_edlevel_distribution_'+self.nickname+'.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Education Level Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype, height=350, x_axis_format="", y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(edlevels)
X = [ x.replace('\ ',' ').replace('$\n$',' ') for x in edlevels.index ] #list(range(nb_element))
Y1 = edlevels.ix[:,'$Non-Certified$'].values
Y2 = edlevels.ix[:,'$Certified$'].values
### Series 1
extra_serie1 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['neutral'],
"format":".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['institute'],
"format":".1f"
}
chart.add_serie(name="Certificate Earners", y=Y2, x=X, extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def age(self,**kwargs):
"""
Creates gender distribution figures.
Parameters (generated during class initialization)
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3',False)
### Removes those users not having the option to fill in edX registration data.
trim_data = self.person[(self.person.registered==1) & (self.person.user_id>156633)]
# Add age column from year_of_birth
trim_data['age'] = trim_data['YoB'].apply(lambda x: datetime.datetime.now().year - x if isinstance(x,int) else None)
age = trim_data.age.dropna()
h,e = np.histogram(age.values,bins=9,range=(0,90))
age = pd.Series(data=h,index=['0-9','10-19','20-29','30-39','40-49','50-59','60-69','70-79','80-89'])
certs = trim_data[trim_data.certified==1]
if certs.username.count() > self.mincerts:
certs = certs.age.dropna()
h,e = np.histogram(certs.values,bins=9,range=(0,90))
certs = pd.Series(data=h,index=['0-9','10-19','20-29','30-39','40-49','50-59','60-69','70-79','80-89'])
else:
certs = pd.Series(index=age.index)
age = pd.concat([age,certs],join='inner',axis=1,keys=['$Non-Certified$','$Certified$'])
age = 100.*age/age.sum()
age = age.apply(lambda x: np.round(x,1))
#print age
#----------------------------------------------------------------
### Static Matplotlib PNG
fig = plt.figure(figsize=(12,6))
ax1 = fig.add_subplot(1,1,1)
age.plot(ax=ax1,kind='bar',color=[xff.colors['neutral'],xff.colors['institute']],rot=40,)
### Plot Details
ax1.set_xticklabels([r'$%s$' % x for x in age.index])
ax1.set_yticklabels([r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],fontsize=30)
ax1.legend(loc=1,prop={'size':28},frameon=False)
### Generalized Plotting functions
figsavename = self.figpath+'age_distribution_'+self.nickname.replace('.','_')
print figsavename
xff.texify(fig,ax1,xlabel='Age',ylabel=None,figsavename=figsavename+'.png')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath+'interactive_age_distribution_'+self.nickname+'.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Age Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype, height=350, x_axis_format="", y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(age)
X = age.index #list(range(nb_element))
Y1 = age['$Non-Certified$'].values
Y2 = age['$Certified$'].values
### Series 1
extra_serie1 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['neutral'],
"format":".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {"tooltip": {"y_start": "", "y_end": "%"},
"color":xff.colors['institute'],
"format":".1f"
}
chart.add_serie(name="Certificate Earners", y=Y2, x=X, extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def scatter_bubble_size(self, DF, colx, coly, disc_act, figsave=False):
"""
Creates scatter plot with x=colx, y=coly.
Size of markers always proportional to disc_act (discussion activity).
Parameters (generated during class initialization)
----------
colx: column to be plotted on x-axis
coly: column to be plotted on y-axis
disc_act: column for scaling marker (bubble) size
figsave: True/False to allow exploratory analysis without saving fig.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
### Data
data = DF[[colx, coly, disc_act, 'certified']].copy()
Jcolx = 0.75
Jcoly = 0.01
bmin = 1.0
bscale = 0.2
data[colx] = data[colx].apply(lambda x: x + Jcolx *
(np.random.sample() - Jcolx))
data[coly] = data[coly].apply(lambda x: x + Jcoly *
(np.random.sample()))
data[disc_act] = data[disc_act].fillna(1.0)
certcut = DF[DF['certified'] == 1].grade.min()
fig = plt.figure(figsize=[12, 10])
ax1 = fig.add_subplot(1, 1, 1)
#Non-Certs
tmp = data[data.certified == 0]
ax1.scatter(tmp[colx],
tmp[coly],
s=bscale * tmp[disc_act],
color=xff.colors['neutral'])
#Certified
tmp = data[data.certified == 1]
ax1.scatter(tmp[colx],
tmp[coly],
s=bscale * tmp[disc_act],
color=xff.colors['institute'])
#ax1.legend(loc=5,prop={'size':18},scatterpoints=1,frameon=False)
ax1.set_xlim(-0.05, )
ax1.set_ylim(-0.05, 1.05)
### Generalized Plotting functions
xff.texify(fig,
ax1,
xlabel=colx,
ylabel=coly,
title='bubble size proportional to %s' %
(disc_act.replace('_', ' ')),
tic_size=20,
label_size=24,
datefontsize=20)
figsavename = self.figpath + 'scatter_' + colx + '_' + coly + '_disc_size_' + self.nickname.replace(
'.', '_') + '.png'
fig.savefig(figsavename, bbox_inches='tight', dpi=300)
return None
def grade_vs_nchapters(self, **kwargs):
"""
Scatter plot of final grade versus nchapters accessed. All points are
jittered for clarity. Text labels are added to indicate subpopulations.
Parameters
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Figures and respective formats.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3', False)
if 'nchapters' not in self.person or 'grade' not in self.person or 'certified' not in self.person:
print "One of the three columns necessary for this plot is missing. Check person_course for: 'nchapters','grade', and 'certified'."
return None
### Data
data = self.person[['nchapters', 'grade', 'certified']].copy()
chap_jmax = 0.75
grade_jmax = 0.005
data.nchapters = data.nchapters.apply(lambda x: x + chap_jmax *
(np.random.sample() - 0.5))
data.grade = data.grade.apply(lambda x: x + grade_jmax *
(np.random.sample()))
data = data.dropna()
certcut = self.person[self.person['certified'] == 1].grade.min()
### Plot
fig = plt.figure(figsize=(12, 10))
ax1 = fig.add_subplot(1, 1, 1)
#Non-Certs
data[data.certified == 0].plot('nchapters',
'grade',
style='.',
color=xff.colors['neutral'],
label=self.nickname,
ax=ax1)
#Certified
data[data.certified == 1].plot('nchapters',
'grade',
style='.',
color=xff.colors['institute'],
ax=ax1)
### Illustrations (labels)
ncmax = self.person[self.person.certified == 1].nchapters.order(
)[-20::].min() ### Funny, but this cuts off staff
ax1.hlines(certcut, 0, ncmax + 1, lw=2)
ax1.vlines(int(ncmax / 2), 0, certcut, lw=2)
ax1.text(ncmax / 6,
certcut - 0.05,
'$Viewed$',
fontsize=30,
alpha=0.75)
ax1.text(ncmax / 2 + (ncmax / 4),
certcut - 0.05,
'$Explored$',
fontsize=30,
alpha=0.75)
ax1.text(
ncmax / 2,
0.8,
'$Certified$',
fontsize=30,
alpha=0.75,
horizontalalignment='center',
)
### Plot Details
ax1.set_xticklabels([r'$%0.f$' % x for x in ax1.get_xticks()])
ax1.set_yticklabels([r'$%0.1f$' % x for x in ax1.get_yticks()],
fontsize=30)
#ax1.legend(loc=4,prop={'size':28},frameon=False)
ax1.set_xlim(0, ncmax + 1)
ax1.set_ylim(0, 1.01)
### Generalized Plotting functions
figsavename = self.figpath + 'scatter_grade_vs_nchapters_' + self.nickname.replace(
'.', '_')
xff.texify(fig,
ax1,
xlabel='Chapters Viewed',
ylabel='Grade',
gridb='y',
figsavename=figsavename + '.png')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
data = data[(data.nchapters > 0) & (data.grade > 0)].dropna()
randrows = np.random.choice(data.index.values, 2000)
data = data.ix[randrows, :]
X1 = data[data.certified == 0].nchapters.values
Y1 = data[data.certified == 0].grade.values
X2 = data[data.certified == 1].nchapters.values
Y2 = data[data.certified == 1].grade.values
#print X1,Y1
### FIGURE
from nvd3 import scatterChart
### Output File
figsavename = self.figpath + 'interactive_scatter_grade_nchap_' + self.nickname + '.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Scatter Plot Grade vs Chapters Viewed: %s" % self._xdata.course_id
chart = scatterChart(name=title,
width=850,
height=550,
x_is_date=False,
x_axis_format=".1f",
y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(X1)
kwargs1 = {'shape': 'circle', 'size': '3'}
kwargs2 = {'shape': 'circle', 'size': '3'}
extra_serie = {"tooltip": {"y_start": "", "y_end": " calls"}}
chart.add_serie(name="Participants",
y=Y1,
x=X1,
extra=extra_serie,
**kwargs1)
chart.add_serie(name="Certified",
y=Y2,
x=X2,
extra=extra_serie,
**kwargs2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def scatter_bubble_size(self,colx,coly,disc_act,figsave=False):
"""
Creates scatter plot with x=colx, y=coly.
Size of markers always proportional to disc_act (discussion activity).
Parameters (generated during class initialization)
----------
colx: column to be plotted on x-axis
coly: column to be plotted on y-axis
disc_act: column for scaling marker (bubble) size
figsave: True/False to allow exploratory analysis without saving fig.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
### Data
data = self.person[[colx,coly,disc_act,'certified']].copy()
Jcolx = 0.75
Jcoly = 0.01
Jcolz = 10 # 1.0/sqrt(10000)
# print Jcolz
bmin = 1.0
bscale = 0.2
data[colx] = data[colx].apply(lambda x: x + Jcolx*(np.random.sample()-Jcolx))
data[coly] = data[coly].apply(lambda x: x + Jcoly*(np.random.sample()))
data[disc_act] = data[disc_act].fillna(1.0).apply(lambda x: x + Jcolz*(np.random.sample()))
### Take top N discussants, and set their dot size to the Nth + 1 highest (lowest of the set)
Nd = 5
topN = data[disc_act].order().index[-Nd:]
data.ix[topN,disc_act] = data.ix[topN[1],disc_act]
if colx=='time_in_course':
data[colx] = data[colx].apply(np.log)
if coly=='time_in_course':
data[coly] = data[coly].apply(np.log)
if disc_act=='time_in_course':
data[disc_act] = data[disc_act].apply(np.log)
certcut = self.person[self.person['certified']==1].grade.min()
fig = plt.figure(figsize=[12,10])
ax1 = fig.add_subplot(1,1,1)
#Non-Certs
tmp = data[data.certified==0]
ax1.scatter(tmp[colx],tmp[coly],s=bscale*tmp[disc_act],color=xff.colors['neutral'])
#Certified
tmp = data[data.certified==1]
ax1.scatter(tmp[colx],tmp[coly],s=bscale*tmp[disc_act],color=xff.colors['institute'])
#ax1.legend(loc=5,prop={'size':18},scatterpoints=1,frameon=False)
ax1.set_ylim(-0.05,1.05)
ax1.set_xlim(6,16)
ax1.set_xticks([np.log(x) for x in [600,3600,10*3600,100*3600]])
ax1.set_xticklabels(['10 min','1 hr','10 hrs','100 hrs'],rotation=40)
# ax1.set_xticks([np.log(x) for x in [1,10,60,600,3600,10*3600,100*3600]])
# ax1.set_xticklabels(['1 sec','10 sec','1 min','10 min','1 hr','10 hrs','100 hrs'],rotation=40)
### Generalized Plotting functions
figsavename = self.figpath+'scatter_'+colx+'_'+coly+'_disc_size_'+self.nickname.replace('.','_')
ylabel = coly.replace('_',' ')
if ylabel == 'time in course':
ylabel = 'Total Time In Course'
xff.texify(fig,ax1,xlabel=colx.replace('_',' '),
ylabel=ylabel,
title='bubble size proportional to %s' % (disc_act.replace('_',' ')),
tic_size=20,label_size=24,datefontsize=20,
figsavename=figsavename+'.png')
return None
def andrew_ho_diagram(self,**kwargs):
"""
Plot showing the intersection of enrollment populations by categories
defined in the 2013 (published 2014) course reports.
http://odl.mit.edu/mitx-working-papers/
Parameters
----------
None
Output
------
Figures and respective formats.
Returns
-------
None
"""
### Registration Types
self.person['Only Registered'] = 0
self.person['Only Viewed'] = 0
self.person['Only Explored'] = 0
g = self.person ### This is a bit silly, but keeps lines short despite the conditions.
### Create disjoint groups (note g and self.person are the same)
#Only Registered: A - (B+C+D)
reg_list = g[ (g['registered']==1) & (g['viewed']==0) & (g['explored']==0) & (g['certified']==0) ].user_id
self.person.ix[self.person[self.person.user_id.isin(reg_list)].index,'Only Registered'] = 1
#Only Viewed: B - (C+D)
view_list = g[ (g.user_id.isin(reg_list)==False) & (g['viewed']==1) & (g['explored']==0) & (g['certified']==0) ].user_id
self.person.ix[self.person[self.person.user_id.isin(view_list)].index,'Only Viewed'] = 1
#Only Explored: C - (D)
exp_list = g[ (g.user_id.isin(reg_list)==False) & (g.user_id.isin(view_list)==False) & (g['explored']==1) & (g['certified']==0) ].user_id
self.person.ix[self.person[self.person.user_id.isin(exp_list)].index,'Only Explored'] = 1
### Figure
fig = plt.figure(figsize=(12,8))
ax1 = fig.add_subplot(111)
#Circles
SCALE = 1000
expratio = SCALE*100.*len(self.person[self.person['Only Explored']==1])/len(self.person[self.person['registered']==1])
#print expratio
if expratio > 17000:
expratio = 1000
certratio = SCALE*100.*len(self.person[self.person['certified']==1])/len(self.person[self.person['registered']==1])
#print certratio
if certratio < 1000:
certratio = 1000
# csize = 1000 # Ratio and csize give relative size of explored and certified circles.
# # expl = len(self.person[self.person['Only Explored']==1])
# # cert = len(self.person[self.person['certified']==1])
ax1.scatter([0.45],[0.5],s=expratio,edgecolor='Black',lw=2,color='white',alpha=0.8)
ax1.scatter([0.50],[0.5],s=certratio,edgecolor=xff.colors['institute'],lw=2,color='white',alpha=0.8)
#Rectangles
rect1 = matplotlib.patches.Rectangle((0,0), 1, 1,fill=False,fc='white',ec='black',lw=2)
rect2 = matplotlib.patches.Rectangle((0.15,0.15), 0.7, 0.7,fill=False,fc='white',ec='black',lw=2)
ax1.add_patch(rect1)
ax1.add_patch(rect2)
FS = 25
### Only registerd
x = len(self.person[self.person['Only Registered']==1])
y = 100.0*x/len(self.person[self.person['registered']==1])
ax1.text(0.05,0.925,'$Only\ registered:\ %d\ \ (%.1f\%%)$' % (x,y),fontsize=FS,ha='left')
x = len(self.person[self.person['Only Viewed']==1])
y = 100.0*x/len(self.person[self.person['registered']==1])
ax1.text(0.2,0.775,'$Only\ viewed:\ %d\ \ (%.1f\%%)$' % (x,y),fontsize=FS,ha='left')
x = len(self.person[self.person['Only Explored']==1])
y = 100.0*x/len(self.person[self.person['registered']==1])
ax1.text(0.175,0.2,'$Only\ explored:$\n$%d\ \ (%.1f\%%)$' % (x,y),fontsize=FS,ha='left')
x = len(self.person[self.person['certified']==1])
y = 100.0*x/len(self.person[self.person['registered']==1])
ax1.text(0.625,0.35,'$Certified:$\n$%d\ \ (%.1f\%%)$' % (x,y),fontsize=FS,ha='left')
ax1.set_xlim([-0.01, 1.01])
ax1.set_ylim([-0.01, 1.01])
## suppress spines
for key in ax1.spines.keys():
ax1.spines[key].set_color('none')
ax1.axes.get_xaxis().set_visible(False)
ax1.axes.get_yaxis().set_visible(False)
### Generalized Plotting functions
figsavename = self.figpath+'AHO_Diagram'+'_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,
figsavename=figsavename+'.png')
# ### Package data
# jsonout = pd.Series()
# for sg in ['Only Registered','Only Viewed','Only Explored','certified']:
# count = len(self.person[self.person[sg]==1].username.unique())
# jsonout.set_value(sg.capitalize()+'(N=%d)'%(count),count)
# jsonout = 100.0*jsonout/jsonout.sum()
# jsondata = []
# for l,v in jsonout.iteritems():
# record = collections.OrderedDict()
# record['label'] = l
# record['value'] = v
# jsondata.append(record)
# print "JSON dump currently commented out."
# # str_jsondata = 'var data = '+json.dumps(jsondata)
# # with open(self.figpath+'percent_participanttypes.json', 'w') as outfile:
# # outfile.write(str_jsondata)
return None
def grade_vs_nchapters(self,**kwargs):
"""
Scatter plot of final grade versus nchapters accessed. All points are
jittered for clarity. Text labels are added to indicate subpopulations.
Parameters
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Figures and respective formats.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3',False)
if 'nchapters' not in self.person or 'grade' not in self.person or 'certified' not in self.person:
print "One of the three columns necessary for this plot is missing. Check person_course for: 'nchapters','grade', and 'certified'."
return None
### Data
data = self.person[['nchapters','grade','certified']].copy()
chap_jmax = 0.75
grade_jmax = 0.005
data.nchapters = data.nchapters.apply(lambda x: x + chap_jmax*(np.random.sample()-0.5))
data.grade = data.grade.apply(lambda x: x + grade_jmax*(np.random.sample()))
data = data.dropna()
certcut = self.person[self.person['certified']==1].grade.min()
### Plot
fig = plt.figure(figsize=(12,10))
ax1 = fig.add_subplot(1,1,1)
#Non-Certs
data[data.certified==0].plot('nchapters','grade',style='.',color=xff.colors['neutral'],label=self.nickname,ax=ax1)
#Certified
data[data.certified==1].plot('nchapters','grade',style='.',color=xff.colors['institute'],ax=ax1)
### Illustrations (labels)
ncmax = self.person[self.person.certified==1].nchapters.order()[-20::].min() ### Funny, but this cuts off staff
ax1.hlines(certcut,0,ncmax+1,lw=2)
ax1.vlines(int(ncmax/2),0,certcut,lw=2)
ax1.text(ncmax/6,certcut-0.05,'$Viewed$',fontsize=30,alpha=0.75)
ax1.text(ncmax/2+(ncmax/4),certcut-0.05,'$Explored$',fontsize=30,alpha=0.75)
ax1.text(ncmax/2,0.8,'$Certified$',fontsize=30,alpha=0.75,horizontalalignment='center',)
### Plot Details
ax1.set_xticklabels([r'$%0.f$' % x for x in ax1.get_xticks()])
ax1.set_yticklabels([r'$%0.1f$' % x for x in ax1.get_yticks()],fontsize=30)
#ax1.legend(loc=4,prop={'size':28},frameon=False)
ax1.set_xlim(0,ncmax+1)
ax1.set_ylim(0,1.01)
### Generalized Plotting functions
figsavename = self.figpath+'scatter_grade_vs_nchapters_'+self.nickname.replace('.','_')
xff.texify(fig,ax1,
xlabel='Chapters Viewed',
ylabel='Grade',
gridb='y',
figsavename=figsavename+'.png')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
data = data[(data.nchapters>0) & (data.grade>0)].dropna()
randrows = np.random.choice(data.index.values,2000)
data = data.ix[randrows,:]
X1 = data[data.certified==0].nchapters.values
Y1 = data[data.certified==0].grade.values
X2 = data[data.certified==1].nchapters.values
Y2 = data[data.certified==1].grade.values
#print X1,Y1
### FIGURE
from nvd3 import scatterChart
### Output File
figsavename = self.figpath+'interactive_scatter_grade_nchap_'+self.nickname+'.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Scatter Plot Grade vs Chapters Viewed: %s" % self._xdata.course_id
chart = scatterChart(name=title, width=850, height=550, x_is_date=False, x_axis_format=".1f", y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(X1)
kwargs1 = {'shape': 'circle', 'size': '3'}
kwargs2 = {'shape': 'circle', 'size': '3'}
extra_serie = {"tooltip": {"y_start": "", "y_end": " calls"}}
chart.add_serie(name="Participants", y=Y1, x=X1, extra=extra_serie, **kwargs1)
chart.add_serie(name="Certified", y=Y2, x=X2, extra=extra_serie, **kwargs2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def content_touches_viz(self, horiz_w_data, certified):
'''
'''
fig = plt.figure(figsize=(32, 12))
fig.subplots_adjust(hspace=.1)
#plt.rcParams.update({'font.size': 20})
ax1 = fig.add_subplot(2, 1, 1)
ax2 = fig.add_subplot(2, 1, 2)
#ax1.plot(tmp['order'],tmp['uniqU'],'o')
bars1 = ax1.bar(horiz_w_data['order'],
horiz_w_data['users'],
3.0,
alpha=0.8,
edgecolor='none')
#Colors the bars (this is where reindexing matters)
for i, b in enumerate(bars1):
flipi = i #len(tmp.index)-1-i
if horiz_w_data.color[flipi] != 'Pink':
bars1[i].set_facecolor(horiz_w_data.color[flipi])
bars1[i].set_edgecolor(horiz_w_data.color[flipi])
else:
bars1[i].set_facecolor('none')
bars1[i].set_edgecolor('none')
#ax1.plot(tmp['order'],tmp.uniqU,'-o',color='Silver',alpha=0.8)
ax1.set_xlabel('Course Structure Index')
ax1.set_ylabel('Unique Users')
#ax1.set_xlim(0,2100)
bars2 = ax2.bar(horiz_w_data['order'],
horiz_w_data.scale,
2.5,
edgecolor='none')
#Colors the bars (this is where reindexing matters)
for i, b in enumerate(bars2):
flipi = i #len(vert.index)-1-i
bars2[i].set_facecolor(horiz_w_data.color[flipi])
invert = True # Choose whether to have the CC plot left or right oriented
ha = 'left'
if invert == True:
ax2.invert_yaxis()
ha = 'right'
ax2.axes.get_xaxis().set_ticks([])
#ax2.set_xlim(0,500)
ax2.axes.get_yaxis().set_ticks([])
#ax2.set_xlim(0,2100)
ax2.set_ylim(7, -1)
#fig.patch.set_visible(False)
ax2.axis('off')
#!!!!! x limits must be set together
ax1.set_xlim(ax2.get_xlim()[0], ax2.get_xlim()[1])
#ax1.set_ylim(0,7500)
xff.texify(fig, ax1, tic_size=32, label_size=32)
dpiset = 300
if certified == True:
figsavename = self.figpath + 'content_touches_horizontal_certified_' + self.nickname.replace(
'.', '_') + '.png'
else:
figsavename = self.figpath + 'content_touches_horizontal_' + self.nickname.replace(
'.', '_') + '.png'
fig.savefig(figsavename, bbox_inches='tight', dpi=dpiset)
return None
def gender(self, **kwargs):
"""
Creates gender distribution figures.
Parameters (generated during class initialization)
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3', False)
### Removes those users not having the option to fill in edX registration data.
trim_data = self.person[(self.person.registered == 1)
& (self.person.user_id > 156633)]
### Data
gdict = {'f': "$Female$", 'm': "$Male$", 'o': "$Other$"}
glist = ['$Female$', '$Male$']
### Munge and Plot
gender = trim_data.gender.dropna().apply(
lambda x: gdict[x]).value_counts()
#print gender
certs = trim_data[trim_data.certified == 1]
if certs.username.count() > self.mincerts:
certs = certs.gender.dropna().apply(
lambda x: gdict[x]).value_counts()
else:
certs = pd.Series(index=gender.index)
gender = pd.concat([gender, certs],
join='inner',
axis=1,
keys=['$Non-Certified$', '$Certified$'])
gender = 100. * gender / gender.sum()
gender = gender.apply(lambda x: np.round(x, 1))
fig = plt.figure(figsize=(12, 6))
ax1 = fig.add_subplot(1, 1, 1)
gender.ix[glist, :].plot(
ax=ax1,
kind='bar',
color=[xff.colors['neutral'], xff.colors['institute']],
rot=0)
### Plot Details
ax1.set_xticklabels([r'%s' % x for x in glist])
ax1.set_yticklabels(
[r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],
fontsize=30)
ax1.legend(loc=2, prop={'size': 28}, frameon=False)
### Generalized Plotting functions
figsavename = self.figpath + 'gender_distribution_' + self.nickname.replace(
'.', '_')
print figsavename
xff.texify(fig,
ax1,
xlabel=None,
ylabel='Count',
figsavename=figsavename + '.png')
# ### Output JSON Records
# gender.name = 'value'
# gender = gender.reset_index().rename(columns={'index':'label'})
# gender.dropna().to_json(figsavename+'.json',orient='records')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
'http://nvd3.org/examples/pie.html'
X = [x.replace('$', '') for x in gender.index]
Y1 = gender.ix[glist, '$Non-Certified$'].values
Y2 = gender.ix[glist, '$Certified$'].values
#----------------------------------------------------------------
### BAR Chart
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath + 'interactive_gender_distribution_' + self.nickname + '.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Gender Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype,
height=350,
x_axis_format="",
y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(gender.ix[glist, :])
### Series 1
extra_serie1 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['neutral'],
"format": ".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['institute'],
"format": ".1f"
}
chart.add_serie(name="Certificate Earners",
y=Y2,
x=X,
extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
#----------------------------------------------------------------
### Pie Chart
from nvd3 import pieChart
### Output File
figsavename = self.figpath + 'interactive_gender_piechart_' + self.nickname + '.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Gender Pie Chart: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = pieChart(name=charttype,
color_category='category20c',
height=400,
width=400)
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
extra_serie = {"tooltip": {"y_start": "", "y_end": " certified"}}
chart.add_serie(y=Y1, x=X, extra=extra_serie)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def scatter_bubble_size(self, colx, coly, disc_act, figsave=False):
"""
Creates scatter plot with x=colx, y=coly.
Size of markers always proportional to disc_act (discussion activity).
Parameters (generated during class initialization)
----------
colx: column to be plotted on x-axis
coly: column to be plotted on y-axis
disc_act: column for scaling marker (bubble) size
figsave: True/False to allow exploratory analysis without saving fig.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
### Data
data = self.person[[colx, coly, disc_act, 'certified']].copy()
Jcolx = 0.75
Jcoly = 0.01
Jcolz = 10 # 1.0/sqrt(10000)
# print Jcolz
bmin = 1.0
bscale = 0.2
data[colx] = data[colx].apply(lambda x: x + Jcolx *
(np.random.sample() - Jcolx))
data[coly] = data[coly].apply(lambda x: x + Jcoly *
(np.random.sample()))
data[disc_act] = data[disc_act].fillna(1.0).apply(lambda x: x + Jcolz *
(np.random.sample()))
### Take top N discussants, and set their dot size to the Nth + 1 highest (lowest of the set)
Nd = 5
topN = data[disc_act].order().index[-Nd:]
data.ix[topN, disc_act] = data.ix[topN[1], disc_act]
if colx == 'time_in_course':
data[colx] = data[colx].apply(np.log)
if coly == 'time_in_course':
data[coly] = data[coly].apply(np.log)
if disc_act == 'time_in_course':
data[disc_act] = data[disc_act].apply(np.log)
certcut = self.person[self.person['certified'] == 1].grade.min()
fig = plt.figure(figsize=[12, 10])
ax1 = fig.add_subplot(1, 1, 1)
#Non-Certs
tmp = data[data.certified == 0]
ax1.scatter(tmp[colx],
tmp[coly],
s=bscale * tmp[disc_act],
color=xff.colors['neutral'])
#Certified
tmp = data[data.certified == 1]
ax1.scatter(tmp[colx],
tmp[coly],
s=bscale * tmp[disc_act],
color=xff.colors['institute'])
#ax1.legend(loc=5,prop={'size':18},scatterpoints=1,frameon=False)
ax1.set_ylim(-0.05, 1.05)
ax1.set_xlim(6, 16)
ax1.set_xticks([np.log(x) for x in [600, 3600, 10 * 3600, 100 * 3600]])
ax1.set_xticklabels(['10 min', '1 hr', '10 hrs', '100 hrs'],
rotation=40)
# ax1.set_xticks([np.log(x) for x in [1,10,60,600,3600,10*3600,100*3600]])
# ax1.set_xticklabels(['1 sec','10 sec','1 min','10 min','1 hr','10 hrs','100 hrs'],rotation=40)
### Generalized Plotting functions
figsavename = self.figpath + 'scatter_' + colx + '_' + coly + '_disc_size_' + self.nickname.replace(
'.', '_')
ylabel = coly.replace('_', ' ')
if ylabel == 'time in course':
ylabel = 'Total Time In Course'
xff.texify(fig,
ax1,
xlabel=colx.replace('_', ' '),
ylabel=ylabel,
title='bubble size proportional to %s' %
(disc_act.replace('_', ' ')),
tic_size=20,
label_size=24,
datefontsize=20,
figsavename=figsavename + '.png')
return None
def age(self, **kwargs):
"""
Creates gender distribution figures.
Parameters (generated during class initialization)
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3', False)
### Removes those users not having the option to fill in edX registration data.
trim_data = self.person[(self.person.registered == 1)
& (self.person.user_id > 156633)]
# Add age column from year_of_birth
trim_data['age'] = trim_data['YoB'].apply(
lambda x: datetime.datetime.now().year - x
if isinstance(x, int) else None)
age = trim_data.age.dropna()
h, e = np.histogram(age.values, bins=9, range=(0, 90))
age = pd.Series(data=h,
index=[
'0-9', '10-19', '20-29', '30-39', '40-49', '50-59',
'60-69', '70-79', '80-89'
])
certs = trim_data[trim_data.certified == 1]
if certs.username.count() > self.mincerts:
certs = certs.age.dropna()
h, e = np.histogram(certs.values, bins=9, range=(0, 90))
certs = pd.Series(data=h,
index=[
'0-9', '10-19', '20-29', '30-39', '40-49',
'50-59', '60-69', '70-79', '80-89'
])
else:
certs = pd.Series(index=age.index)
age = pd.concat([age, certs],
join='inner',
axis=1,
keys=['$Non-Certified$', '$Certified$'])
age = 100. * age / age.sum()
age = age.apply(lambda x: np.round(x, 1))
#print age
#----------------------------------------------------------------
### Static Matplotlib PNG
fig = plt.figure(figsize=(12, 6))
ax1 = fig.add_subplot(1, 1, 1)
age.plot(
ax=ax1,
kind='bar',
color=[xff.colors['neutral'], xff.colors['institute']],
rot=40,
)
### Plot Details
ax1.set_xticklabels([r'$%s$' % x for x in age.index])
ax1.set_yticklabels(
[r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],
fontsize=30)
ax1.legend(loc=1, prop={'size': 28}, frameon=False)
### Generalized Plotting functions
figsavename = self.figpath + 'age_distribution_' + self.nickname.replace(
'.', '_')
print figsavename
xff.texify(fig,
ax1,
xlabel='Age',
ylabel=None,
figsavename=figsavename + '.png')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath + 'interactive_age_distribution_' + self.nickname + '.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Age Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype,
height=350,
x_axis_format="",
y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(age)
X = age.index #list(range(nb_element))
Y1 = age['$Non-Certified$'].values
Y2 = age['$Certified$'].values
### Series 1
extra_serie1 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['neutral'],
"format": ".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['institute'],
"format": ".1f"
}
chart.add_serie(name="Certificate Earners",
y=Y2,
x=X,
extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def level_of_education(self, **kwargs):
'''
Plot Level of Education Attained; typically taken from the edX enrollment questionairre.
'''
"""
Creates distribution of highest level of education attained.
Parameters (generated during class initialization)
----------
NVD3 = False. If true, nvd3 interactive figure output.
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
NVD3 = kwargs.get('NVD3', False)
### Level of Education (LoE)
### Data
eddict = {
'el': "Less\ than$\n$ Secondary",
'jhs': "Less\ than$\n$ Secondary",
'none': "Less\ than$\n$ Secondary",
'hs': "Secondary",
'a': "Secondary",
'b': "Bachelor\'s",
'm': "Master\'s",
'p_se': "Doctorate",
'p_oth': "Doctorate",
'p': "Doctorate",
'other': None,
'NA': None,
'nan': None,
}
edlist = [
"Less\ than$\n$ Secondary", "Secondary", "Bachelor\'s",
"Master\'s", "Doctorate"
]
trim_data = self.person[(self.person.registered == 1)
& (self.person.user_id > 156633)]
edlevels = trim_data.LoE.apply(lambda x: eddict[str(x)] if x in eddict.
keys() else None).value_counts()[edlist]
if trim_data[trim_data.certified ==
1].username.count() > self.mincerts:
certs = trim_data[trim_data.certified == 1].LoE.apply(
lambda x: eddict[str(x)]
if x in eddict.keys() else None).value_counts()[edlist]
else:
certs = pd.Series(index=edlevels.index)
edlevels = pd.concat([edlevels, certs],
join='inner',
axis=1,
keys=['$Non-Certified$', '$Certified$'])
edlevels = 100. * edlevels / edlevels.sum()
edlevels = edlevels.apply(lambda x: np.round(x, 1))
#print edlevels
#Plot
fig = plt.figure(figsize=(12, 6))
ax1 = fig.add_subplot(1, 1, 1)
edlevels.plot(ax=ax1,
kind='bar',
color=[xff.colors['neutral'], xff.colors['institute']],
rot=40)
### Plot Details
ax1.set_xticklabels([r'$%s$' % x for x in edlist])
ax1.set_yticklabels(
[r'${0}\%$'.format("%.0f" % (y)) for y in ax1.get_yticks()],
fontsize=30)
ax1.legend(loc=2, prop={'size': 22}, frameon=False)
### Generalized Plotting functions
figsavename = self.figpath + 'loe_distribution_' + self.nickname.replace(
'.', '_')
print figsavename
xff.texify(fig,
ax1,
xlabel=None,
ylabel=None,
figsavename=figsavename + '.png')
### Output JSON Records
#cc.name = 'value'
#cc = cc.reset_index().rename(columns={'index':'label'})
#cc.dropna().to_json(figsavename+'.json',orient='records')
#----------------------------------------------------------------
### NVD3 Interactive http://nvd3.org/
if NVD3:
### FIGURE
from nvd3 import multiBarChart
### Output File
figsavename = self.figpath + 'interactive_edlevel_distribution_' + self.nickname + '.html'
output_file = open(figsavename, 'w')
print figsavename
title = "Education Level Distribution: %s" % self._xd.course_id
charttype = 'multiBarChart'
chart = multiBarChart(name=charttype,
height=350,
x_axis_format="",
y_axis_format=".1f")
chart.set_containerheader("\n\n<h2>" + title + "</h2>\n\n")
nb_element = len(edlevels)
X = [
x.replace('\ ', ' ').replace('$\n$', ' ')
for x in edlevels.index
] #list(range(nb_element))
Y1 = edlevels.ix[:, '$Non-Certified$'].values
Y2 = edlevels.ix[:, '$Certified$'].values
### Series 1
extra_serie1 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['neutral'],
"format": ".1f"
}
chart.add_serie(name="Participants", y=Y1, x=X, extra=extra_serie1)
### Series 2
extra_serie2 = {
"tooltip": {
"y_start": "",
"y_end": "%"
},
"color": xff.colors['institute'],
"format": ".1f"
}
chart.add_serie(name="Certificate Earners",
y=Y2,
x=X,
extra=extra_serie2)
### Final Output
chart.buildhtml()
output_file.write(chart.htmlcontent)
#---------------------------------------
#close Html file
output_file.close()
return None
def daily_activity(self):
"""
Creates daily timeseries of discussion activity (only posts/comments/votes from forum data).
Parameters (generated during class initialization)
----------
None
Output
------
Saves figures to specified directories.
Returns
-------
None
"""
fig = plt.figure(figsize=[20, 6])
ax1 = fig.add_subplot(1, 1, 1)
### List of Certified Participants
certs = self.pc_plus[self.pc_plus.certified == 1].user_id.unique()
certs = [str(u) for u in certs]
### Non-Certified
post_act = self.forum[(self.forum.created_at.notnull()) & (
self.forum.author_id.isin(certs) == False)].created_at.apply(
lambda x: x.date()).value_counts().sort_index()
post_act.plot(ax=ax1,
style="-o",
ms=6,
lw=2,
color=xff.colors['neutral'],
label='$Non-Certified$')
#(post_act.cumsum()/10).plot(ax=ax1,style="-",ms=3,color='Orange')
### Certified
post_act = self.forum[(self.forum.created_at.notnull()) & (
self.forum.author_id.isin(certs))].created_at.apply(
lambda x: x.date()).value_counts().sort_index()
post_act.plot(ax=ax1,
style="-o",
ms=6,
lw=2,
color=xff.colors['institute'],
label='$Certified$')
xmin = (self.cinfo['start_date'] -
np.timedelta64(2, 'W')).item().date()
xmax = (self.cinfo['end_date'] + np.timedelta64(4, 'W')).item().date()
ax1.set_xlim(xmin, xmax)
ylim1 = ax1.get_ylim()[1]
ax1.vlines([
self.cinfo.start_date.item().date(),
self.cinfo.end_date.item().date()
],
0,
ylim1,
colors='Gray',
lw=1.5,
linestyles='--')
ax1.set_ylim(0, ylim1)
ax1 = xff.timeseries_plot_formatter(ax1, interval=1)
ax1.legend(loc=1, prop={'size': 24}, frameon=False)
figsavename = self.figpath + 'discussion_activity_' + self.nickname.replace(
'.', '_')
xff.texify(fig,
ax1,
ylabel='Forum Text Submissions',
tic_size=20,
label_size=24,
datefontsize=20,
title=self.nickname,
figsavename=figsavename + '.png')
return None