-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathshap_analysis.py
383 lines (286 loc) · 14.7 KB
/
shap_analysis.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
#!/usr/bin/env python
# coding: utf-8
# In[1]:
import pandas as pd
import numpy as np
from matplotlib import pyplot as plt
from datetime import datetime
from math import sqrt
from numpy import concatenate
from sklearn.preprocessing import MinMaxScaler
from sklearn.preprocessing import LabelEncoder
from sklearn.metrics import mean_squared_error
from sklearn.metrics import mean_absolute_error
import tensorflow
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout
from tensorflow.keras.layers import LSTM
from tensorflow.compat.v1.keras.backend import get_session
tensorflow.compat.v1.disable_v2_behavior()
from __future__ import print_function
import shap
shap.initjs()
# In[2]:
# import date and time information and parse it into Pandas DataFrame index
def parse(x):
return datetime.strptime(x, '%Y %m')
filename = 'procedure_imported_data/TB_LASSO_index17.csv'
dataset = pd.read_csv(filename, engine='python', parse_dates=[['year', 'month']],
index_col=0, date_parser=parse)
dataset.drop('No', axis=1, inplace=True)
# manually specify column names
dataset.columns = ['rate', 'avr_air_pressure', 'avr_wind_spd', 'rel_humidity',
'month_sun', 'hour_sun', 'extreme_wind_speed', 'min_air_pressure',
'max_temp', 'day_max_precipitation', 'power', 'industrial_add',
'sales_rate','fin_budget', 'cpi', 'pas_turn', 'im_ex',
'ex_factory_index']
dataset.index.name = 'date'
# save as processed data file
completed_filename = 'completed_data/completed_TB_LASSO_index17.csv'
dataset.to_csv(completed_filename)
# In[3]:
# data settings
dataset = pd.read_csv(completed_filename, engine='python', header=0, index_col=0)
values = dataset.values
groups = [0, 1, 2, 3, 4, 5, 6, 7]
i = 1
plt.figure()
for group in groups:
plt.subplot(len(groups), 1, i)
plt.plot(values[:, group])
plt.title(dataset.columns[group], y=0.3, loc='right')
i += 1
plt.show()
# In[4]:
# convert data sets
def series_to_supervised(data, n_in=1, n_out=1, dropnan=True):
n_vars = 1 if type(data) is list else data.shape[1]
df = pd.DataFrame(data)
cols, names = list(), list()
# input sequences (t-n, ... t-1)
for i in range(n_in, 0, -1):
cols.append(df.shift(i))
names += [('var%d(t-%d)' % (j+1, i)) for j in range(n_vars)]
# predict sequences (t, t+1, ... t+n)
for i in range(0, n_out):
cols.append(df.shift(-i))
if i == 0:
names += [('var%d(t)' % (j+1)) for j in range(n_vars)]
else:
names += [('var%d(t+%d)' % (j+1, i)) for j in range(n_vars)]
# information summary
agg = pd.concat(cols, axis=1)
agg.columns = names
# delete rows with 'NAN'
if dropnan:
agg.dropna(inplace=True)
return agg
# data setup
new_filename = 'completed_data/completed_TB_LASSO_index17.csv'
dataset = pd.read_csv(new_filename, engine='python', header=0, index_col=0)
values = dataset.values
values = values.astype('float32')
# data normalization process (Min-Max method)
scaler = MinMaxScaler(feature_range=(0, 1))
scaled = scaler.fit_transform(values)
reframed = series_to_supervised(scaled, 1, 1)
reframed.drop(reframed.columns[[19, 20, 21, 22, 23, 24, 25,
26, 27, 28, 29, 30, 31, 32,
33, 34, 35]], axis=1, inplace=True)
print(reframed.head())
# In[5]:
# split into training and testing sets
values = reframed.values
n_train_hours = 11 * 12 - 1
train = values[:n_train_hours, :]
test = values[n_train_hours:, :]
# split into input and output
train_X, train_y = train[:, :-1], train[:, -1]
test_X, test_y = test[:, :-1], test[:, -1]
# reshape input to 3D [sample, steps, feature]
train_X = train_X.reshape((train_X.shape[0], 1, train_X.shape[1]))
test_X = test_X.reshape((test_X.shape[0], 1, test_X.shape[1]))
print(train_X.shape, train_y.shape, test_X.shape, test_y.shape)
# In[6]:
# design neural network (add network structure layer by layer)
model = Sequential()
model.add(LSTM(8, input_shape=(train_X.shape[1], train_X.shape[2])))
model.add(Dense(1))
model.compile(loss='mae', optimizer='adam')
model.fit(train_X, train_y, epochs=150, batch_size=22, validation_data=(test_X, test_y))
# In[7]:
import shap
# we use the first 132 training examples as our background dataset to integrate over
explainer = shap.DeepExplainer(model, train_X[:132])
# explain the first 24 predictions
# explaining each prediction requires 2 * background dataset size runs
shap_values = explainer.shap_values(test_X[:24])
# In[8]:
dataset.columns = ['rate', 'avr_air_pressure', 'avr_wind_spd', 'rel_humidity',
'month_sun', 'hour_sun', 'extreme_wind_speed', 'min_air_pressure',
'max_temp', 'day_max_precipitation', 'power', 'industrial_add',
'sales_rate','fin_budget', 'cpi', 'pas_turn', 'im_ex',
'ex_factory_index']
fig1 = shap.force_plot(explainer.expected_value[0], shap_values[0][0], dataset.columns,
matplotlib=True, show=False)
fig1.set_facecolor('white')
fig1.savefig('single sample feature value.pdf', bbox_inches = 'tight', dpi=300)
# In[9]:
# set SHAP value manually
new_shape_values = [[-2.07555909e-02, 3.22832377e-03, -7.01126968e-03,
-3.28778382e-03, 2.44493019e-02, -3.09205391e-02,
1.08185103e-02, -7.04229064e-03, 4.86812964e-02,
-5.11698658e-03, -1.55109155e-04, -7.64225982e-03,
1.74246449e-03, -2.45297775e-02, -1.45195471e-03,
5.29442768e-05, -4.70874831e-03, 4.39360924e-03],
[-6.59693871e-03, 5.19968895e-03, -3.33166984e-03,
2.58122142e-02, -2.40170844e-02, -8.21811147e-03,
1.16919205e-02, -1.06574930e-02, 5.94768934e-02,
-4.27860953e-03, -2.52412912e-03, -4.75753890e-03,
8.64983420e-04, -2.24942714e-02, -1.97302387e-03,
-2.61467770e-02, 1.78980641e-03, 3.35772289e-03],
[-6.18082983e-03, 2.19015777e-03, -3.23731406e-03,
2.56482549e-02, -1.80464834e-02, -5.58605557e-03,
-9.41686856e-04, -4.36894735e-03, 4.13419195e-02,
-2.78691365e-03, -3.22899059e-03, -5.42938430e-03,
4.60640236e-04, -3.29626985e-02, -1.33975944e-03,
-1.69023033e-02, 3.29275406e-03, 4.00972459e-03],
[1.87410903e-03, 8.17836029e-04, 4.30113170e-03,
3.27303447e-02, -2.48822849e-02, 1.66430101e-02,
-2.60072004e-04, -2.99071497e-03, 2.04449077e-03,
-3.28044430e-03, -5.10160904e-03, -6.70398492e-03,
-8.65057751e-04, -3.33607048e-02, -1.37072662e-03,
-1.92170106e-02, 6.97918353e-04, 4.21113428e-03],
[-1.73768355e-03, -1.03258190e-03, 8.47992115e-03,
3.15088369e-02, -1.67861581e-02, 2.01348569e-02,
-7.00937677e-03, 3.28904833e-03, -1.07546030e-02,
-1.37725170e-03, -4.93122777e-03, -8.34029447e-03,
8.97017657e-04, -2.61410046e-02, -1.64264359e-03,
-1.57973617e-02, 7.40339048e-04, 2.65506003e-03],
[-8.15949403e-03, -1.35604758e-04, 1.06013082e-02,
2.00399160e-02, -1.02066407e-02, 1.30257793e-02,
-1.60173886e-02, 1.20612206e-02, -2.72354130e-02,
6.46398403e-03, -6.41402975e-03, -7.89591298e-03,
5.86519483e-04, -2.53124405e-02, -1.86213164e-03,
-5.91297681e-03, 5.74062346e-04, 2.14769086e-03],
[-9.77965351e-03, -1.39463739e-03, -2.80151796e-03,
-6.05971599e-03, 1.72903370e-02, 1.19558431e-03,
2.35291664e-03, 4.16230364e-03, -3.17660533e-02,
1.42680923e-03, -8.35874118e-03, -6.03611115e-03,
-1.37069833e-03, -3.33966836e-02, -1.71921239e-03,
-4.23008390e-03, 4.95605520e-04, 1.70980697e-03],
[-1.29434448e-02, -1.59713102e-03, -6.54948363e-03,
-2.54679695e-02, 4.30313870e-02, -1.12912226e-02,
4.27989289e-03, 2.74793641e-03, -3.82700078e-02,
6.74198521e-03, -1.02390414e-02, -5.51288016e-03,
-9.26182547e-04, -3.23635489e-02, -2.23907758e-03,
2.37068089e-05, 6.73407165e-04, 1.28452328e-03],
[-1.22452145e-02, -2.26697163e-03, -7.00565660e-03,
-2.35434789e-02, 1.14653260e-02, 7.30588520e-03,
4.86261584e-03, 8.75421427e-03, -4.07758839e-02,
5.26203401e-03, -5.66830207e-03, -6.07494032e-03,
2.39119702e-03, -3.32556702e-02, -2.57747411e-03,
7.82727636e-03, 5.29345241e-04, 4.97613393e-04],
[-1.55314347e-02, -4.62123338e-04, -1.07944207e-02,
-2.66383588e-02, 4.10183705e-02, -1.84664484e-02,
1.68761087e-03, 7.59532861e-03, -2.10026950e-02,
-2.93137832e-03, -3.37803806e-03, -8.44099186e-03,
1.50625396e-03, -3.14089134e-02, -1.51637266e-03,
3.87080805e-03, -1.66192430e-03, -2.94341648e-04],
[-1.70360319e-02, 4.98240290e-04, -9.27673362e-04,
2.05317592e-05, 9.22270771e-03, -2.01043189e-02,
3.07649164e-03, -3.30243446e-03, -1.59196574e-02,
-2.92849960e-03, -4.18327795e-03, -1.01872161e-02,
6.61080005e-03, -2.00458448e-02, -2.16689217e-03,
-1.92473433e-03, 1.74601935e-03, -7.53690954e-04],
[-1.45101165e-02, 1.08633481e-03, -1.50221819e-03,
4.36263764e-03, 1.32262977e-02, -2.99540758e-02,
1.71765895e-03, -1.53972756e-03, 2.22219639e-02,
-3.92881036e-03, -3.53255495e-03, -8.73056985e-03,
4.83861612e-03, -1.84153374e-02, -2.33037700e-03,
-1.73530597e-02, -9.42786864e-04, -1.40932598e-03],
[-2.79291067e-02, 7.75961846e-04, -1.67368678e-03,
-1.79934106e-03, 2.11611260e-02, -3.87369059e-02,
3.54862795e-03, -5.74171869e-03, 4.11593914e-02,
-4.18859301e-03, -3.72431614e-03, -6.70281518e-03,
6.99011190e-03, -2.82287542e-02, -2.44962587e-03,
-1.84938032e-02, -2.05021747e-03, -2.22685840e-03],
[-6.11664215e-03, 1.45033293e-03, -6.54179184e-03,
1.52186584e-02, 1.62637164e-03, -2.08775662e-02,
1.99746131e-03, -8.25796556e-03, 5.03733344e-02,
-4.99503873e-03, -1.02623156e-03, -5.12921624e-03,
4.52045671e-04, 1.23958895e-03, -1.04366045e-03,
-2.20232792e-02, -1.89182849e-03, -3.21546686e-03],
[-4.23329696e-03, 1.92260952e-03, 9.85658960e-04,
2.65115462e-02, -1.64232182e-03, -1.87600087e-02,
1.93231343e-03, -1.16155995e-03, 4.19396609e-02,
-3.50433961e-03, -2.19895854e-03, -5.51427156e-03,
4.80194518e-04, -3.06371646e-03, -3.52524081e-03,
-1.74480211e-02, 7.47543163e-05, -3.88583937e-03],
[1.36704103e-03, 1.02080661e-03, -2.10598484e-03,
3.77250127e-02, -1.69778746e-02, -9.19245882e-04,
5.23353787e-03, -5.50091127e-03, 1.65197290e-02,
-3.93319456e-03, -5.34757320e-03, -5.28483139e-03,
-5.04200207e-03, -6.14599884e-03, -3.99033213e-03,
-2.17563212e-02, -1.47264136e-03, -4.28709062e-03],
[-1.46247272e-03, -1.57702365e-03, 5.89598436e-03,
4.24824916e-02, -2.32027341e-02, 2.45868731e-02,
-1.99274998e-03, 6.46002404e-03, -2.88530346e-02,
-2.15467322e-03, -3.52529669e-03, -5.71665075e-03,
4.69931541e-03, -1.03305485e-02, -3.49102938e-03,
-1.86834298e-02, -7.60147639e-04, -3.62309557e-03],
[1.86585370e-04, -6.89294015e-04, 9.52232257e-03,
2.70879734e-02, -3.03317178e-02, 3.50458361e-02,
-1.25583373e-02, 3.36104725e-03, -3.60212997e-02,
-1.76215626e-03, -2.97341635e-03, -7.32240221e-03,
6.16864534e-03, -1.00793121e-02, -3.44743649e-03,
-7.28188921e-03, -2.35085539e-03, -3.22515005e-03],
[-1.07350331e-02, -2.03957246e-03, -3.96988774e-03,
1.02152126e-02, -1.50731821e-02, 2.12810840e-02,
2.53679376e-04, 2.54714140e-03, -5.03333621e-02,
1.20046898e-03, -7.99648836e-03, -4.08936990e-03,
-6.89459266e-04, -1.07963961e-02, -3.38995433e-03,
-7.04406621e-03, -2.44290772e-04, -3.19778593e-03],
[-8.81627481e-03, -5.99310908e-04, -3.19746253e-03,
-2.12441739e-02, 2.48745736e-02, -9.74084716e-04,
-3.58108082e-03, 4.70031146e-03, -4.43474799e-02,
5.66944899e-03, -1.00030638e-02, -4.24292358e-03,
-8.77651037e-04, -9.22032073e-03, -3.60374269e-03,
-2.24848860e-03, -2.09333119e-03, -3.07639269e-03],
[-7.73078622e-03, -2.68002436e-03, -1.05096018e-02,
-3.20781022e-02, 3.10305823e-02, -4.61669732e-03,
8.09112447e-04, 5.10913925e-03, -3.67721468e-02,
1.74991805e-02, -2.90492317e-03, -2.32674438e-03,
5.80509752e-03, -1.02832066e-02, -3.58918519e-03,
5.82328904e-03, -1.56172365e-03, -3.22169112e-03],
[-1.51959890e-02, -4.45540092e-04, -7.39430636e-03,
-1.62583906e-02, 4.82000643e-03, 4.79325978e-03,
-2.32608733e-03, 3.76569486e-04, -2.67435256e-02,
-1.13612576e-03, -3.03718843e-03, -2.84173666e-03,
1.99531252e-03, -8.29647109e-03, -2.45142239e-03,
2.19182950e-03, -2.72843824e-03, -3.97775462e-03],
[-9.14163049e-03, 1.12673210e-03, -2.98561342e-03,
6.67838566e-03, -3.77204083e-03, -5.44741424e-03,
1.56690471e-03, -5.67982532e-03, -6.40143594e-03,
-7.05961138e-05, -3.29612545e-03, -1.65848038e-03,
4.80439048e-03, -9.86603182e-03, -2.25911639e-03,
-4.17489093e-03, -2.51824036e-04, -4.15934063e-03],
[-1.14836525e-02, 7.47481827e-04, 5.05208829e-03,
3.30525227e-02, -1.39971273e-02, -7.80009851e-03,
-4.08052187e-03, -6.22018706e-03, 1.27807977e-02,
-2.58079614e-03, -6.71766652e-03, -1.89971889e-03,
4.37216973e-03, -9.97895095e-03, -2.22787494e-03,
-2.13507023e-02, -3.15412902e-03, -3.50207952e-03]]
new_shape_values = np.array(new_shape_values)
print(new_shape_values)
# In[13]:
filename = 'procedure_imported_data/shap_test.csv'
X = pd.read_csv(filename, engine='python')
print(shap_values)
shap.summary_plot(new_shape_values[:24, :], X.iloc[:24, :], show=False)
plt.savefig('Feature density scatter plot.pdf', bbox_inches = 'tight', dpi=600)
# In[11]:
shap.force_plot(explainer.expected_value, new_shape_values[:24, :], X.iloc[:24, :])
# In[15]:
shap.summary_plot(new_shape_values[:24, :], X.iloc[:24,:], plot_type="bar", show=False)
plt.savefig('Feature importance SHAP value.pdf', bbox_inches = 'tight', dpi=600)