def get_line(datas, user_id):
#datas按照时间逆序排序，即最近的在前面
if os.path.exists("static/img/" + user_id + "/" + "line.png"):
return "static/img/" + user_id + "/" + "line.png"
num = len(datas) > 7 and 7 or len(datas)
x = list(range(1, num + 1))
#分三条线
scene = [0] * 7
person = [0] * 7
animals = [0] * 7
#从数据开始进行筛选
for i in range(num):
if datas[i][0] == '场景':
scene[i] = random.random() * 5
elif datas[i][0] == '人物':
person[i] = random.random() * 5
else:
animals[i] = random.random() * 5
plt.title('Result Analysis')
plt.plot(x, scene, color='green', label='scene')
plt.plot(x, person, color='blue', label='person')
plt.plot(x, animals, color='red', label='animals')
plt.set_size_inches(3, 2)
plt.legend()
# 显示图例
plt.xlabel('latest videl')
plt.ylabel('rate')
plt.savefig("static/img/" + user_id + "/" + "line.png")
return "static/img/" + user_id + "/" + "line.png"
def define_strategy(X_train, y_train):
"""
Identifies the best strategy for implementing the dummy classifier
Parameters:
----------
X_train: Subset of the original data set to be used as training set
y_train: Labels corresponding to the X_train data set
"""
plt = matplotlib.pyplot.gcf()
plt.set_size_inches(12, 8)
strats = ['stratified', 'most_frequent', 'prior', 'uniform', 'constant']
train_dummy_scores = {}
for clfs in strats:
if clfs == 'constant':
dummy_clf = DummyClassifier(strategy = clfs, random_state = 0, \
constant = 0)
else:
dummy_clf = DummyClassifier(strategy=clfs, random_state=0)
dummy_clf.fit(X_train, y_train)
score = dummy_clf.score(X_train, y_train)
train_dummy_scores[clfs] = score
values = list(train_dummy_scores.values())
ax = sns.stripplot(strats, values)
ax.set(xlabel='strategy', ylabel='training score')
plt.savefig('../plots/strategies.jpg')
plt.clf()
def draw_wavelet_transform():
from scipy.interpolate import interp1d
import scipy.fftpack
from scipy import signal
sensor_data = data = np.genfromtxt('./ICS_slipperData/Alice0105db.csv', dtype=float, delimiter=',', names=True)
timestamp = sensor_data['Timestamp']
signal = sensor_data['Axis1']
sample_frequency = 500
timestamp_sampled = np.arange(timestamp[0], timestamp[-1], 1.0/sample_frequency)
f = interp1d(timestamp, signal, kind='slinear')
signal_sample = f(timestamp_sampled)
widths = np.arange(1, 200)
cwtmatr = scipy.signal.cwt(scipy.fftpack.fft(f(timestamp_sampled)), scipy.signal.ricker, widths)
print len(timestamp_sampled)
print cwtmatr.shape
plt.figure(figsize=(16, 6), dpi=100)
plt.set_size_inches(8.27, 11.69)
#plt.xticks(timestamp_sampled)
plt.imshow(cwtmatr)
#plt.show()
plt.savefig('signal_cwt.png', dpi=600, bbox_inches='tight')
def multiplotter(df, leftdict={}, rightdict={}, **kwargs):
"""
Plot a big chart and its subplots together
:param leftdict: a dict of arguments for the big plot
:type leftdict: dict
:param rightdict: a dict of arguments for the small plot
:type rightdict: dict
"""
from .table import Table
axes = []
df2 = rightdict.pop('data', df.copy())
# add more cool layouts here
# and figure out a nice way to access them other than numbers...
from .layouts import layouts
if isinstance(kwargs.get('layout'), list):
layout = kwargs['layout']
else:
lay = kwargs.get('layout', 1)
layout = layouts.get(lay)
kinda = leftdict.pop('kind', 'area')
kindb = rightdict.pop('kind', 'line')
tpa = leftdict.pop('transpose', False)
tpb = rightdict.pop('transpose', False)
numtoplot = leftdict.pop('num_to_plot', len(layout) - 1)
ntpb = rightdict.pop('num_to_plot', 'all')
sharex = rightdict.pop('sharex', True)
sharey = rightdict.pop('sharey', False)
if kindb == 'pie':
piecol = rightdict.pop('colours', 'default')
coloursb = rightdict.pop('colours', 'default')
fig = plt.figure()
for i, (nrows, ncols, plot_number) in enumerate(layout, start=-1):
if tpb:
df2 = df2.T
if i >= len(df2.columns):
continue
ax = fig.add_subplot(nrows, ncols, plot_number)
if i == -1:
df.chart(kind=kinda, ax=ax, transpose=tpa, num_to_plot=numtoplot, **leftdict)
if kinda in ['bar', 'hist', 'barh']:
import matplotlib as mpl
colmap = []
rects = [r for r in ax.get_children() if isinstance(r, mpl.patches.Rectangle)]
for r in rects:
if r._facecolor not in colmap:
colmap.append(r._facecolor)
try:
colmap = {list(df.columns)[i]: x for i, x in enumerate(colmap[:len(df.columns)])}
except AttributeError:
colmap = {list(df.index)[i]: x for i, x in enumerate(colmap[:len(df.index)])}
else:
try:
colmap = {list(df.columns)[i]: l.get_color() for i, l in enumerate(ax.get_lines())}
except AttributeError:
colmap = {list(df.index)[i]: l.get_color() for i, l in enumerate(ax.get_lines())}
else:
if colmap and kindb != 'pie':
try:
name = df2.iloc[:, i].name
coloursb = colmap[name]
except IndexError:
coloursb = 'gray'
except KeyError:
coloursb = 'gray'
if kindb == 'pie':
rightdict['colours'] = piecol
else:
rightdict['colours'] = coloursb
Table(df2.iloc[:, i]).chart(kind=kindb, ax=ax, sharex=sharex, sharey=sharey,
num_to_plot=ntpb, **rightdict)
ax.set_title(df2.iloc[:, i].name)
wspace = kwargs.get('wspace', .2)
hspace = kwargs.get('hspace', .5)
fig.subplots_adjust(bottom=0.025, left=0.025, top=0.975, right=0.975, wspace=wspace, hspace=hspace)
size = kwargs.get('figsize', (10, 4))
fig.set_size_inches(size)
try:
plt.set_size_inches(size)
except:
pass
if kwargs.get('save'):
savepath = os.path.join('images', kwargs['save'] + '.png')
fig.savefig(savepath, dpi=150)
return plt
FangShui=['Bawal','Chennai','Roorkee','T-16 Auto','Auto']
dcf5=pd.DataFrame()
for i in range(5):
if i==4:
abhikeliye=df.loc[FangShui[i]:,:]
else:
abhikeliye=df.loc[FangShui[i]:FangShui[i+1],:]
abhikeliye=abhikeliye.iloc[7,:]
abhikeliye=abhikeliye.iloc[:48]
abhikeliye=abhikeliye.astype(str).astype(float).apply(np.int64)
abhikeliye.index = pd.date_range(start='4/1/2015', periods=len(data), freq='M')
abhikeliye.plot()
plt.ylabel("Productivity per Person per plant", fontsize=18)
plt.xlabel("Months", fontsize=18)
plt.tight_layout()
plt.set_size_inches(18.5, 10.5, forward=True)
if 'pdf' in option:
plt.savefig('download/foo4.pdf', bbox_inches='tight',dpi=200)
if 'eps' in option:
plt.savefig('download/foo4.eps', bbox_inches='tight',dpi=200)
if 'png' in option:
plt.savefig('download/foo4.png', bbox_inches='tight', dpi=200)
dcf5=pd.concat([dcf5,abhikeliye], axis=1)
dcf5.plot()
plt.legend(labels=FangShui,loc=2, prop={'size': 5})
plt.ylabel("Productivity per Person per plant", fontsize=11)
plt.xlabel("Months", fontsize=11)
option=['png','pdf','eps']
if 'pdf' in option:
plt.savefig('download/hoo4.pdf', bbox_inches='tight',dpi=200)
df_july.rolling(5).mean().plot(ax=ax)
df_aug.rolling(5).mean().plot(ax=ax)
plt.grid()
plt.legend(['NAO May', 'NAO June', 'NAO July', 'NAO August'])
ax = df_JJA.rolling(5).mean().plot()
df_JJ.rolling(5).mean().plot(ax=ax)
plt.grid()
plt.legend(['NAO JJA', 'NAO JJ'])
ax = df_JJ.plot(color='k', linewidth=.5, zorder=10)
ax.fill_between([df_JJ.index[0], df_JJ.index[-1]], 0, [2, 2], color='steelblue', alpha=.4)
ax.fill_between([df_JJ.index[0], df_JJ.index[-1]], 0, [-2, -2], color='indianred', alpha=.4)
df_JJ.rolling(5).mean().plot(ax=ax, color='dimgray', linewidth=3, zorder=10)
plt.grid()
plt.title('NAO June-July')
plt.ylim([-1.5, 1.5])
plt.xlabel(None)
ax.get_legend().remove()
plt.set_size_inches(w=15, h=10)
plt.savefig('NAO_JJ.png', dpi=200)
# Save csv
df_JJ.index = df_JJ.index.year
df_JJ.to_csv('NAO_June-July.csv', float_format='%.2f')
df_JJ.rolling(5).mean().to_csv('NAO_June-July_5yr_rolling_mean.csv', float_format='%.2f')
def multiplotter(df, leftdict={},rightdict={}, **kwargs):
"""
Plot a big chart and its subplots together
:param leftdict: a dict of arguments for the big plot
:type leftdict: dict
:param rightdict: a dict of arguments for the small plot
:type rightdict: dict
"""
from corpkit.interrogation import Interrogation
import matplotlib.pyplot as plt
axes = []
if isinstance(df, Interrogation):
df = df.results
df2 = rightdict.pop('data', df.copy())
if isinstance(df2, Interrogation):
df2 = df2.results
# add more cool layouts here
# and figure out a nice way to access them other than numbers...
from corpkit.layouts import layouts
if isinstance(kwargs.get('layout'), list):
layout = kwargs.get('layout')
else:
lay = kwargs.get('layout', 1)
layout = layouts.get(lay)
kinda = leftdict.pop('kind', 'area')
kindb = rightdict.pop('kind', 'line')
tpa = leftdict.pop('transpose', False)
tpb = rightdict.pop('transpose', False)
numtoplot = leftdict.pop('num_to_plot', len(layout) - 1)
ntpb = rightdict.pop('num_to_plot', 'all')
sharex = rightdict.pop('sharex', True)
sharey = rightdict.pop('sharey', False)
if kindb == 'pie':
piecol = rightdict.pop('colours', 'default')
coloursb = rightdict.pop('colours', 'default')
fig = plt.figure()
for i, (nrows, ncols, plot_number) in enumerate(layout, start=-1):
if tpb:
df2 = df2.T
if i >= len(df2.columns):
continue
ax = fig.add_subplot(nrows, ncols, plot_number)
if i == -1:
df.visualise(kind=kinda, ax=ax, transpose=tpa,
num_to_plot=numtoplot, **leftdict)
if kinda in ['bar', 'hist', 'barh']:
import matplotlib as mpl
colmap = []
rects = [r for r in ax.get_children() if isinstance(r, mpl.patches.Rectangle)]
for r in rects:
if r._facecolor not in colmap:
colmap.append(r._facecolor)
colmap = {list(df.columns)[i]: x for i, x in enumerate(colmap[:len(df.columns)])}
else:
colmap = {list(df.columns)[i]: l.get_color() for i, l in enumerate(ax.get_lines())}
else:
if colmap and kindb != 'pie':
try:
name = df2.iloc[:, i].name
coloursb = colmap[name]
except IndexError:
coloursb = 'gray'
except KeyError:
coloursb = 'gray'
if kindb == 'pie':
rightdict['colours'] = piecol
else:
rightdict['colours'] = coloursb
df2.iloc[:, i].visualise(kind=kindb, ax=ax, sharex=sharex, sharey=sharey,
num_to_plot=ntpb, **rightdict)
ax.set_title(df2.iloc[:, i].name)
wspace = kwargs.get('wspace', .2)
hspace = kwargs.get('hspace', .5)
fig.subplots_adjust(bottom=0.025, left=0.025, top=0.975, right=0.975,
wspace=wspace, hspace=hspace)
size = kwargs.get('figsize', (10, 4))
fig.set_size_inches(size)
try:
plt.set_size_inches(size)
except:
pass
if kwargs.get('save'):
import os
from corpkit.process import urlify
savepath = os.path.join('images', urlify(kwargs['save']) + '.png')
fig.savefig(savepath, dpi=150)
return plt