def make_category_scatter_figure():
categories = get_data()
# Build a figure and place a single axes instance in it.
fig = plt.figure(figsize=(4,3))
ax1 = fig.add_subplot(1,1,1)
# Use our helper function to do the category scatter plot.
cs = CategoryScatter( ax1, categories )
# Locate the axes spines on the left and bottom.
spine_placer(ax1, location='left,bottom' )
# Finalize the category scatter plot (stuff that can only be done
# after the spines are placed).
cs.finalize()
# Add standard y label.
ax1.set_ylabel('y value (units)')
# Now, add a final few touchups.
ax1.spines['bottom'].set_color('none') # don't draw bottom spine
ax1.yaxis.set_major_locator( mticker.MaxNLocator(nbins=4) )
auto_reduce_spine_bounds( ax1 )
fig.tight_layout()
return fig
def plot_ci_for_ms(path, figname, vals, figsize=(6,4)):
figname = os.path.splitext(figname)[0]
NAMES = {'TH>trpA1 head':'TH>trpA1 (head)',
'TH>trpA1 thorax':'TH>trpA1 (thorax)',
'TH head':'TH (head)',
'TH thorax':'TH (thorax)',
'trpA1 head':'trpA1 (head)',
'trpA1 thorax':'trpA1 (thorax)',
'controls head':'controls (head)',
'controls thorax':'controls (thorax)',
}
COLORS = {'TH>trpA1 head':flymad_plot.RED,
'TH>trpA1 thorax':flymad_plot.ORANGE,
'TH head':flymad_plot.GREEN,
'TH thorax':flymad_plot.BLUE,
'trpA1 head':flymad_plot.GREEN,
'trpA1 thorax':flymad_plot.BLUE,
'controls head':flymad_plot.GREEN,
'controls thorax':flymad_plot.BLUE,
}
ORDER = ['TH>trpA1 head',
'TH>trpA1 thorax',
'TH head',
'TH thorax',
'trpA1 head',
'trpA1 thorax',
'controls head',
'controls thorax',
]
figure_title = "THGAL4 %s cumulative incidence" % figname
fig = plt.figure(figsize=figsize)
ax = fig.add_subplot(1,1,1)
for gt in sorted(vals, cmp=lambda a,b: cmp(ORDER.index(a), ORDER.index(b))):
ax.plot(vals[gt]['x'],vals[gt]['y'],
lw=2,clip_on=False,
color=COLORS[gt],label=NAMES[gt])
spine_placer(ax, location='left,bottom' )
ax.legend()
ax.set_xlabel('Time (s)')
ax.set_ylabel('Cumulative incidence (%)')
ax.set_ylim([0,100])
ax.set_xlim([0,20])
flymad_plot.retick_relabel_axis(ax, [0,10,20], [0,100])
fig.savefig(flymad_plot.get_plotpath(path,"thgal4_ci_%s.png" % figname), bbox_inches='tight')
fig.savefig(flymad_plot.get_plotpath(path,"thgal4_ci_%s.svg" % figname), bbox_inches='tight')
def make_many_timeseries_figure():
# Generate some fake data in 4 timeseries
times = np.arange( 0, 200.0, 0.01 )
nt = len(times)
ys = []
for i in range(4):
ys.append( np.sin( times*2*np.pi*0.02 + i*20 ) + 0.13*np.random.randn(nt) )
# Build a figure and place a two axes instances in it.
fig = plt.figure(figsize=(8,6))
ax1 = fig.add_subplot(2,1,1)
ax2 = fig.add_subplot(2,1,2)
# Use our helper function to plot the multiple timeseries.
mts1 = ManyTimeseries( ax1, times, ys, show=['all','mean'] )
# Use our helper function to plot the multiple timeseries.
mts2 = ManyTimeseries( ax2, times, ys, show=['mean','std'] )
# Locate the axes spines.
spine_placer(ax1, location='left' )
spine_placer(ax2, location='left,bottom' )
# Finalize the plots (stuff that can only be done
# after the spines are placed).
mts1.finalize()
mts2.finalize()
# Add standard axis labels.
ax1.set_ylabel('y1 value (units)')
ax2.set_ylabel('y2 value (units)')
ax2.set_xlabel('time (seconds)')
# Now, add a final few touchups.
ax1.yaxis.set_major_locator( mticker.MaxNLocator(nbins=4) )
ax1.set_xticks([])
ax2.yaxis.set_major_locator( mticker.MaxNLocator(nbins=4) )
auto_reduce_spine_bounds( ax1 )
auto_reduce_spine_bounds( ax2 )
fig.tight_layout()
return fig
def plot_data(path, data):
ci_html_buf = ''
for exp_name in data:
gts = data[exp_name].keys()
laser = '130ht'
gts_string = 'vs'.join(gts)
figname = laser + '_' + gts_string
fig = plt.figure("Song (%s)" % figname, figsize=(10, 6))
ax = fig.add_subplot(1, 1, 1)
datasets = {}
for gt in gts:
if flymad_analysis.genotype_is_exp(gt):
color = flymad_plot.RED
order = 1
elif flymad_analysis.genotype_is_ctrl(gt):
color = flymad_plot.BLACK
order = 2
elif flymad_analysis.genotype_is_trp_ctrl(gt):
order = 3
color = flymad_plot.BLUE
else:
color = 'cyan'
order = 0
gtdf = data[exp_name][gt]
datasets[gt] = dict(xaxis=gtdf['mean']['t'].values,
value=gtdf['mean']['zx'].values,
std=gtdf['std']['zx'].values,
n=gtdf['n']['zx'].values,
order=order,
df=gtdf,
label=flymad_analysis.human_label(gt),
color=color,
N=len(gtdf['df']['obj_id'].unique()))
pvalue_buf = ''
for gt in datasets:
label = flymad_analysis.human_label(gt)
if '>' not in label:
continue
# OK, this is a Gal4 + UAS - do head vs thorax stats
gtdf = data[exp_name][gt]['df']
ci_data = do_cum_incidence(gtdf, label)
ci_html_buf += ci_data['buf']
ax_cum = ci_data['ax']
spine_placer(ax_cum, location='left,bottom')
ax_cum.set_ylabel('Fraction extending wing (%)')
ax_cum.set_xlabel('Time (s)')
note = '%s\n%s\np-value: %.3g\n%d flies\nn=%d, %d' % (
label, madplot.p2stars(ci_data['p_value']), ci_data['p_value'],
len(gtdf['obj_id'].unique()), ci_data['n_head'],
ci_data['n_thorax'])
ax_cum.text(
0,
1, #top left
note,
fontsize=10,
horizontalalignment='left',
verticalalignment='top',
transform=ax_cum.transAxes,
zorder=0)
ci_data['fig'].savefig(flymad_plot.get_plotpath(
path, "song_cum_inc_%s.png" % figname),
bbox_inches='tight')
ci_data['fig'].savefig(flymad_plot.get_plotpath(
path, "song_cum_inc_%s.svg" % figname),
bbox_inches='tight')
# for i in range(len(head_times)):
# head_values = gtdf[gtdf['t']==head_times[i]]
# thorax_values = gtdf[gtdf['t']==thorax_times[i]]
# test1 = head_values['zx'].values
# test2 = thorax_values['zx'].values
# hval, pval = kruskal(test1, test2)
# pvalue_buf += 'Pulse %d: Head vs thorax WEI p-value: %.3g (n=%d, %d)\n'%(
# i+1, pval, len(test1), len(test2) )
#all experiments used identical activation times
headtargetbetween = dict(
xaxis=data['pIP10']['wtrpmyc']['first']['t'].values,
where=data['pIP10']['wtrpmyc']['first']['ttm'].values > 0)
thoraxtargetbetween = dict(
xaxis=data['pIP10']['wtrpmyc']['first']['t'].values,
where=data['pIP10']['wtrpmyc']['first']['ttm'].values < 0)
flymad_plot.plot_timeseries_with_activation(
ax,
targetbetween=[headtargetbetween, thoraxtargetbetween],
sem=True,
note=pvalue_buf,
**datasets)
ax.set_xlabel('Time (s)')
ax.set_ylabel('Wing extension index')
ax.set_ylim([-0.05, 0.4] if gts_string ==
"40347vswtrpmycvs40347trpmyc" else [-0.05, 1.0])
ax.set_xlim([-10, 180])
flymad_plot.retick_relabel_axis(
ax, [0, 60, 120, 180], [0, 0.2, 0.4]
if gts_string == "40347vswtrpmycvs40347trpmyc" else [0, 0.5, 1])
fig.savefig(flymad_plot.get_plotpath(path, "song_%s.png" % figname),
bbox_inches='tight')
fig.savefig(flymad_plot.get_plotpath(path, "song_%s.svg" % figname),
bbox_inches='tight')
with open(flymad_plot.get_plotpath(path, "song_cum_in.html"),
mode='w') as fd:
fd.write(ci_html_buf)
def plot_ci_for_ms(path, figname, vals, figsize=(6, 4)):
figname = os.path.splitext(figname)[0]
NAMES = {
'TH>trpA1 head': 'TH>trpA1 (head)',
'TH>trpA1 thorax': 'TH>trpA1 (thorax)',
'TH head': 'TH (head)',
'TH thorax': 'TH (thorax)',
'trpA1 head': 'trpA1 (head)',
'trpA1 thorax': 'trpA1 (thorax)',
'controls head': 'controls (head)',
'controls thorax': 'controls (thorax)',
}
COLORS = {
'TH>trpA1 head': flymad_plot.RED,
'TH>trpA1 thorax': flymad_plot.ORANGE,
'TH head': flymad_plot.GREEN,
'TH thorax': flymad_plot.BLUE,
'trpA1 head': flymad_plot.GREEN,
'trpA1 thorax': flymad_plot.BLUE,
'controls head': flymad_plot.GREEN,
'controls thorax': flymad_plot.BLUE,
}
ORDER = [
'TH>trpA1 head',
'TH>trpA1 thorax',
'TH head',
'TH thorax',
'trpA1 head',
'trpA1 thorax',
'controls head',
'controls thorax',
]
figure_title = "THGAL4 %s cumulative incidence" % figname
fig = plt.figure(figsize=figsize)
ax = fig.add_subplot(1, 1, 1)
for gt in sorted(vals,
cmp=lambda a, b: cmp(ORDER.index(a), ORDER.index(b))):
ax.plot(vals[gt]['x'],
vals[gt]['y'],
lw=2,
clip_on=False,
color=COLORS[gt],
label=NAMES[gt])
spine_placer(ax, location='left,bottom')
ax.legend()
ax.set_xlabel('Time (s)')
ax.set_ylabel('Cumulative incidence (%)')
ax.set_ylim([0, 100])
ax.set_xlim([0, 20])
flymad_plot.retick_relabel_axis(ax, [0, 10, 20], [0, 100])
fig.savefig(flymad_plot.get_plotpath(path, "thgal4_ci_%s.png" % figname),
bbox_inches='tight')
fig.savefig(flymad_plot.get_plotpath(path, "thgal4_ci_%s.svg" % figname),
bbox_inches='tight')
color='black',
linewidth=0.5,
)
ax.errorbar(
[this_x_value],
[np.mean(this_y_values)],
[scipy.stats.sem(this_y_values)],
clip_on=False,
color='black',
linewidth=0.5,
capsize=2,
)
xticks.append(this_x_value)
tick_labels.append(name_key)
spine_placer(ax, location='left,bottom')
ax.spines['bottom'].set_color('none')
ax.spines['bottom'].set_position(('outward', 5))
ax.set_yticks([0, 20])
ax.set_xticks(xticks)
ax.set_xticklabels(tick_labels, rotation='vertical')
ax.set_ylabel('latency (sec)')
fig1.subplots_adjust(left=0.4, bottom=0.65) # do not clip text
svg1_fname = 'th_gal4_latency_%s_%s.svg' % (measurement,
pooled_str)
fig1.savefig(svg1_fname)
# --- plots --------------------------------
fig2 = plt.figure('scatter ' + measurement + pooled_str,
figsize=(3, 5))
ax = fig2.add_subplot(111)
clip_on=False,
color='black',
linewidth=0.5,
)
ax.errorbar( [this_x_value],
[ np.mean(this_y_values) ],
[ scipy.stats.sem( this_y_values ) ],
clip_on=False,
color='black',
linewidth=0.5,
capsize=2,
)
xticks.append( this_x_value )
tick_labels.append( name_key )
spine_placer(ax, location='left,bottom' )
ax.spines['bottom'].set_color('none')
ax.spines['bottom'].set_position(('outward',5))
ax.set_yticks([0,20])
ax.set_xticks(xticks)
ax.set_xticklabels(tick_labels, rotation='vertical')
ax.set_ylabel('latency (sec)')
fig1.subplots_adjust(left=0.4,bottom=0.65) # do not clip text
svg1_fname = 'th_gal4_latency_%s_%s.svg'%(measurement,pooled_str)
fig1.savefig(svg1_fname)
# --- plots --------------------------------
fig2 = plt.figure('scatter '+measurement+pooled_str,figsize=(3,5))
ax = fig2.add_subplot(111)
tick_labels = []
xticks = []
def plot_data(path, data):
ci_html_buf = ''
for exp_name in data:
gts = data[exp_name].keys()
laser = '130ht'
gts_string = 'vs'.join(gts)
figname = laser + '_' + gts_string
fig = plt.figure("Song (%s)" % figname,figsize=(10,6))
ax = fig.add_subplot(1,1,1)
datasets = {}
for gt in gts:
if flymad_analysis.genotype_is_exp(gt):
color = flymad_plot.RED
order = 1
elif flymad_analysis.genotype_is_ctrl(gt):
color = flymad_plot.BLACK
order = 2
elif flymad_analysis.genotype_is_trp_ctrl(gt):
order = 3
color = flymad_plot.BLUE
else:
color = 'cyan'
order = 0
gtdf = data[exp_name][gt]
datasets[gt] = dict(xaxis=gtdf['mean']['t'].values,
value=gtdf['mean']['zx'].values,
std=gtdf['std']['zx'].values,
n=gtdf['n']['zx'].values,
order=order,
df=gtdf,
label=flymad_analysis.human_label(gt),
color=color,
N=len(gtdf['df']['obj_id'].unique()))
pvalue_buf = ''
for gt in datasets:
label=flymad_analysis.human_label(gt)
if '>' not in label:
continue
# OK, this is a Gal4 + UAS - do head vs thorax stats
gtdf = data[exp_name][gt]['df']
ci_data = do_cum_incidence(gtdf, label)
ci_html_buf += ci_data['buf']
ax_cum = ci_data['ax']
spine_placer(ax_cum, location='left,bottom' )
ax_cum.set_ylabel('Fraction extending wing (%)')
ax_cum.set_xlabel('Time (s)')
note = '%s\n%s\np-value: %.3g\n%d flies\nn=%d, %d'%(label,
madplot.p2stars(ci_data['p_value']),
ci_data['p_value'],
len(gtdf['obj_id'].unique()),
ci_data['n_head'],
ci_data['n_thorax'])
ax_cum.text(0, 1, #top left
note,
fontsize=10,
horizontalalignment='left',
verticalalignment='top',
transform=ax_cum.transAxes,
zorder=0)
ci_data['fig'].savefig(flymad_plot.get_plotpath(path,"song_cum_inc_%s.png" % figname),
bbox_inches='tight')
ci_data['fig'].savefig(flymad_plot.get_plotpath(path,"song_cum_inc_%s.svg" % figname),
bbox_inches='tight')
# for i in range(len(head_times)):
# head_values = gtdf[gtdf['t']==head_times[i]]
# thorax_values = gtdf[gtdf['t']==thorax_times[i]]
# test1 = head_values['zx'].values
# test2 = thorax_values['zx'].values
# hval, pval = kruskal(test1, test2)
# pvalue_buf += 'Pulse %d: Head vs thorax WEI p-value: %.3g (n=%d, %d)\n'%(
# i+1, pval, len(test1), len(test2) )
#all experiments used identical activation times
headtargetbetween = dict(xaxis=data['pIP10']['wtrpmyc']['first']['t'].values,
where=data['pIP10']['wtrpmyc']['first']['ttm'].values > 0)
thoraxtargetbetween = dict(xaxis=data['pIP10']['wtrpmyc']['first']['t'].values,
where=data['pIP10']['wtrpmyc']['first']['ttm'].values < 0)
flymad_plot.plot_timeseries_with_activation(ax,
targetbetween=[headtargetbetween,thoraxtargetbetween],
sem=True,
note=pvalue_buf,
**datasets
)
ax.set_xlabel('Time (s)')
ax.set_ylabel('Wing extension index')
ax.set_ylim([-0.05,0.4] if gts_string == "40347vswtrpmycvs40347trpmyc" else [-0.05,1.0])
ax.set_xlim([-10,180])
flymad_plot.retick_relabel_axis(ax, [0, 60, 120, 180],
[0, 0.2, 0.4] if gts_string == "40347vswtrpmycvs40347trpmyc" else [0, 0.5, 1])
fig.savefig(flymad_plot.get_plotpath(path,"song_%s.png" % figname), bbox_inches='tight')
fig.savefig(flymad_plot.get_plotpath(path,"song_%s.svg" % figname), bbox_inches='tight')
with open(flymad_plot.get_plotpath(path,"song_cum_in.html"),mode='w') as fd:
fd.write(ci_html_buf)
def plot_timeseries_with_activation(ax, targetbetween=None, downsample=1, sem=False,
legend_location='upper right', note="",
individual=None, individual_title=None,
marker=None,linestyle='-',markersize=1,
return_dict=False,
**datasets):
ORDER_LAST = 100
DEFAULT_COLORS = {"exp":RED,"ctrl":BLACK}
trans = mtransforms.blended_transform_factory(ax.transData, ax.transAxes)
final_copy = os.environ.get('FLYMAD_FINAL')
def _ds(a):
if downsample == 1:
return a
else:
tmp = []
for i in range(0,len(a),downsample):
vals = a[i:i+downsample]
tmp.append( scipy.stats.nanmean(vals) )
return np.array(tmp)
def _dn(a,b):
nans_a, = np.where(np.isnan(a))
nans_b, = np.where(np.isnan(b))
nans_all = np.union1d(nans_a, nans_b)
n_nans = len(nans_all)
if n_nans:
print "\tremoving %d nans from plot" % n_nans
clean_a = np.delete(a,nans_all)
clean_b = np.delete(b,nans_all)
return clean_a, clean_b
def _sort_by_order(a,b):
return cmp(datasets[a].get('order', ORDER_LAST), datasets[b].get('order', ORDER_LAST))
def _fill_between(f_ax, f_xaxis, f_where, f_facecolor, f_zorder):
f_ax.fill_between(f_xaxis, 0, 1, where=f_where,
edgecolor='none', facecolor=f_facecolor,
alpha=0.15, transform=trans, zorder=f_zorder)
if targetbetween is not None:
if not (isinstance(targetbetween, list) or isinstance(targetbetween, tuple)):
targetbetween = [targetbetween]
for tb in targetbetween:
_fill_between(ax, tb['xaxis'], tb['where'], tb.get('facecolor','yellow'),
tb.get('zorder',1))
if any(['std' in datasets[exp] for exp in datasets]):
note += "+/- SEM\n" if sem else "+/- STD\n"
note += "" if downsample == 1 else ("downsample x %d\n" % downsample)
#zorder = 1 = back
top_zorder = 60
bottom_zorder = 30
plotted = {}
exps_colors = [plt.cm.gnuplot(i) for i in np.linspace(0, 1.0, len(datasets))]
cur_zorder = 2
for data,default_color in zip(sorted(datasets.keys(), cmp=_sort_by_order), exps_colors):
exp = datasets[data]
label = exp.get('label',data)
note += "N(%s)=%s\n" % (label,exp.get('N','??'))
if exp.get('ontop'):
this_zorder = top_zorder + cur_zorder
else:
this_zorder = bottom_zorder + cur_zorder
print "plotting %s (%s) zorder %s" % (label,data,this_zorder)
if 'std' in exp:
if sem:
spread = exp['std'] / np.sqrt(exp['n'])
else:
spread = exp['std']
else:
spread = None
if exp.get('color') is None:
color = default_color
else:
color = DEFAULT_COLORS.get(data,default_color)
if spread is not None:
ax.fill_between(exp['xaxis'][::downsample], _ds(exp['value']+spread), _ds(exp['value']-spread),
alpha=0.1, color=color,
zorder=this_zorder)
x,y = _dn(exp['xaxis'][::downsample], _ds(exp['value']))
zorder = this_zorder + 1
plotted[data] = dict(x=x,y=y,color=color,label=label,zorder=zorder)
ax.plot(x,y,
color=color,label=label,
lw=2,linestyle=linestyle,clip_on=exp.get('clip_on',True),
marker=marker,markerfacecolor=color,markersize=markersize,markeredgecolor='none',
zorder=zorder)
cur_zorder -= 2
spine_placer(ax, location='left,bottom' )
l = ax.legend(loc=legend_location)
l.set_zorder(1+top_zorder+cur_zorder)
ax.text(0, 1, #top left
note,
fontsize=10,
horizontalalignment='left',
verticalalignment='top',
transform=ax.transAxes,
color='white' if final_copy else 'k',
zorder=-100)
axs = [ax]
figs = {}
if (not final_copy) and (individual is not None) and isinstance(individual, dict):
for data in individual:
try:
#try to avoid creating many duplicate figures
#based on the title be unique if provided
if individual_title:
fignum = hash(individual_title)
if plt.fignum_exists(fignum):
continue
else:
fignum = None
individual_title = ""
individual_title += data
gdf = datasets[data]['df']
groupcol = individual[data]['groupby']
xcol = individual[data]['xaxis']
ycol = individual[data]['yaxis']
grouper = gdf.groupby(groupcol)
nts = len(grouper)
gs,nr_nc = get_gridspec_to_fit_nplots(nts)
fig2 = plt.figure(fignum, figsize=(2*max(nr_nc),2*max(nr_nc)))
fig2.suptitle(individual_title)
print 'plotting', nts, 'individual timeseries for', data
for gridspec_id,(name,group) in zip(gs,grouper):
#we can't assume these are numpy arrays here, but
#np.array does the right thing and converts pandas things
#while being no-op on np arrays
grp_xaxis = np.array(group[xcol])
grp_yaxis = np.array(group[ycol])
iax = fig2.add_subplot(gridspec_id)
print "\tplot",name,xcol,"vs",ycol
x,y = _dn(grp_xaxis[::downsample], _ds(grp_yaxis))
iax.plot(x, y,
label=str(name), color='k')
iax.legend(loc=legend_location)
# if fb_wherecol:
# _fill_between(iax,
# grp_xaxis,
# np.array(group[fb_wherecol]) > 0,
# 'yellow')
axs.append(iax)
figs[md5.md5(individual_title).hexdigest()] = fig2
except KeyError, e:
print "\terror plotting individual timeseries (%s)" % e
def plot_timeseries_with_activation(ax,
targetbetween=None,
downsample=1,
sem=False,
legend_location='upper right',
note="",
individual=None,
individual_title=None,
marker=None,
linestyle='-',
markersize=1,
return_dict=False,
**datasets):
ORDER_LAST = 100
DEFAULT_COLORS = {"exp": RED, "ctrl": BLACK}
trans = mtransforms.blended_transform_factory(ax.transData, ax.transAxes)
final_copy = os.environ.get('FLYMAD_FINAL')
def _ds(a):
if downsample == 1:
return a
else:
tmp = []
for i in range(0, len(a), downsample):
vals = a[i:i + downsample]
tmp.append(scipy.stats.nanmean(vals))
return np.array(tmp)
def _dn(a, b):
nans_a, = np.where(np.isnan(a))
nans_b, = np.where(np.isnan(b))
nans_all = np.union1d(nans_a, nans_b)
n_nans = len(nans_all)
if n_nans:
print "\tremoving %d nans from plot" % n_nans
clean_a = np.delete(a, nans_all)
clean_b = np.delete(b, nans_all)
return clean_a, clean_b
def _sort_by_order(a, b):
return cmp(datasets[a].get('order', ORDER_LAST),
datasets[b].get('order', ORDER_LAST))
def _fill_between(f_ax, f_xaxis, f_where, f_facecolor, f_zorder):
f_ax.fill_between(f_xaxis,
0,
1,
where=f_where,
edgecolor='none',
facecolor=f_facecolor,
alpha=0.15,
transform=trans,
zorder=f_zorder)
if targetbetween is not None:
if not (isinstance(targetbetween, list)
or isinstance(targetbetween, tuple)):
targetbetween = [targetbetween]
for tb in targetbetween:
_fill_between(ax, tb['xaxis'], tb['where'],
tb.get('facecolor', 'yellow'), tb.get('zorder', 1))
if any(['std' in datasets[exp] for exp in datasets]):
note += "+/- SEM\n" if sem else "+/- STD\n"
note += "" if downsample == 1 else ("downsample x %d\n" % downsample)
#zorder = 1 = back
top_zorder = 60
bottom_zorder = 30
plotted = {}
exps_colors = [
plt.cm.gnuplot(i) for i in np.linspace(0, 1.0, len(datasets))
]
cur_zorder = 2
for data, default_color in zip(sorted(datasets.keys(), cmp=_sort_by_order),
exps_colors):
exp = datasets[data]
label = exp.get('label', data)
note += "N(%s)=%s\n" % (label, exp.get('N', '??'))
if exp.get('ontop'):
this_zorder = top_zorder + cur_zorder
else:
this_zorder = bottom_zorder + cur_zorder
print "plotting %s (%s) zorder %s" % (label, data, this_zorder)
if 'std' in exp:
if sem:
spread = exp['std'] / np.sqrt(exp['n'])
else:
spread = exp['std']
else:
spread = None
if exp.get('color') is None:
color = default_color
else:
color = DEFAULT_COLORS.get(data, default_color)
if spread is not None:
ax.fill_between(exp['xaxis'][::downsample],
_ds(exp['value'] + spread),
_ds(exp['value'] - spread),
alpha=0.1,
color=color,
zorder=this_zorder)
x, y = _dn(exp['xaxis'][::downsample], _ds(exp['value']))
zorder = this_zorder + 1
plotted[data] = dict(x=x, y=y, color=color, label=label, zorder=zorder)
ax.plot(x,
y,
color=color,
label=label,
lw=2,
linestyle=linestyle,
clip_on=exp.get('clip_on', True),
marker=marker,
markerfacecolor=color,
markersize=markersize,
markeredgecolor='none',
zorder=zorder)
cur_zorder -= 2
spine_placer(ax, location='left,bottom')
l = ax.legend(loc=legend_location)
l.set_zorder(1 + top_zorder + cur_zorder)
ax.text(
0,
1, #top left
note,
fontsize=10,
horizontalalignment='left',
verticalalignment='top',
transform=ax.transAxes,
color='white' if final_copy else 'k',
zorder=-100)
axs = [ax]
figs = {}
if (not final_copy) and (individual is not None) and isinstance(
individual, dict):
for data in individual:
try:
#try to avoid creating many duplicate figures
#based on the title be unique if provided
if individual_title:
fignum = hash(individual_title)
if plt.fignum_exists(fignum):
continue
else:
fignum = None
individual_title = ""
individual_title += data
gdf = datasets[data]['df']
groupcol = individual[data]['groupby']
xcol = individual[data]['xaxis']
ycol = individual[data]['yaxis']
grouper = gdf.groupby(groupcol)
nts = len(grouper)
gs, nr_nc = get_gridspec_to_fit_nplots(nts)
fig2 = plt.figure(fignum,
figsize=(2 * max(nr_nc), 2 * max(nr_nc)))
fig2.suptitle(individual_title)
print 'plotting', nts, 'individual timeseries for', data
for gridspec_id, (name, group) in zip(gs, grouper):
#we can't assume these are numpy arrays here, but
#np.array does the right thing and converts pandas things
#while being no-op on np arrays
grp_xaxis = np.array(group[xcol])
grp_yaxis = np.array(group[ycol])
iax = fig2.add_subplot(gridspec_id)
print "\tplot", name, xcol, "vs", ycol
x, y = _dn(grp_xaxis[::downsample], _ds(grp_yaxis))
iax.plot(x, y, label=str(name), color='k')
iax.legend(loc=legend_location)
# if fb_wherecol:
# _fill_between(iax,
# grp_xaxis,
# np.array(group[fb_wherecol]) > 0,
# 'yellow')
axs.append(iax)
figs[md5.md5(individual_title).hexdigest()] = fig2
except KeyError, e:
print "\terror plotting individual timeseries (%s)" % e
