-
Notifications
You must be signed in to change notification settings - Fork 0
/
ElectionsDataPreperation.py
358 lines (291 loc) · 16.7 KB
/
ElectionsDataPreperation.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
import numpy as np
import Consts
import pandas as pd
import os
from pandas import read_csv, to_numeric, Series
from sklearn.model_selection import train_test_split
from matplotlib import pyplot as plt
class ElectionsDataPreperation:
""" main class that is used for data preperation
"""
def __init__(self, sInputFileTrain, sInputFileVal, sInputFileTest, sInputFileTrainLabel, sInputFileValLabel,
sInputFileTestLabel):
# fields that contain path to X files of train, val and test
self.sInputFileTrain = sInputFileTrain
self.sInputFileVal = sInputFileVal
self.sInputFileTest = sInputFileTest
# fields that contain path to y files of train, val, test
self.sInputFileTrainLabel = sInputFileTrainLabel
self.sInputFileValLabel = sInputFileValLabel
self.sInputFileTestLabel = sInputFileTestLabel
# fields that contain x and Y for train, test and validation
self.trainData = None
self.trainLabels = None
self.valData = None
self.valLabels = None
self.testData = None
self.testLabels = None
def save_labels(self, base: str, set: int):
header = [Consts.VOTE_STR, Consts.INDEX_COL]
df = self.trainLabels.loc[:, header]
df.to_csv(base.format(set, Consts.FileSubNames.Y_TRAIN.value), header=header, index=False)
df = self.valLabels.loc[:, header]
df.to_csv(base.format(set, Consts.FileSubNames.Y_VAL.value), header=header, index=False)
df = self.testLabels.loc[:, header]
df.to_csv(base.format(set, Consts.FileSubNames.Y_TEST.value), header=header, index=False)
def save_data(self, base: str, set: int):
self.trainData.to_csv(base.format(set, Consts.FileSubNames.X_TRAIN.value), index=False)
self.valData.to_csv(base.format(set, Consts.FileSubNames.X_VAL.value), index=False)
self.testData.to_csv(base.format(set, Consts.FileSubNames.X_TEST.value), index=False)
def loadAndImpute(self, lDataTypes=None):
""" lDataTypes is a list with following values ['test', 'validation'], the list determines if the data load,
renaming and imputation will happen on the test and validation sets
"""
self.loadData(lDataTypes)
self._changeStringToValues(lDataTypes)
self._changeVoteToNumber(lDataTypes)
# first we impute train
self._dataImpute(self.trainData, self.trainData, self.sInputFileTrain)
if Consts.DataTypes.TEST in lDataTypes:
self._dataImpute(self.trainData, self.testData, self.sInputFileTest)
if Consts.DataTypes.VAL in lDataTypes:
self._dataImpute(self.trainData, self.valData, self.sInputFileVal)
def loadData(self, lDataTypes = list(Consts.DataTypes)):
"""lDataTypes is a list with following values ['test', 'validation'], the list determines if the data load,
renaming and imputation will happen on the test and validation sets
"""
trainFileNameX = self.sInputFileTrain
self.trainData = read_csv(trainFileNameX, header=0, keep_default_na=True)
trainFileNameY = self.sInputFileTrainLabel
self.trainLabels = read_csv(trainFileNameY, header=0)
if (Consts.DataTypes.TEST in lDataTypes):
testFileNameX = self.sInputFileTest
self.testData = read_csv(testFileNameX, header=0, keep_default_na=True)
testFileNameY = self.sInputFileTestLabel
self.testLabels = read_csv(testFileNameY, header=0)
if (Consts.DataTypes.VAL in lDataTypes):
valFileNameX = self.sInputFileVal
self.valData = read_csv(valFileNameX, header=0, keep_default_na=True)
valFileNameY = self.sInputFileValLabel
self.valLabels = read_csv(valFileNameY, header=0)
def fix_nan_and_outliers(self):
self._dataImpute(self.trainData, self.trainData, self.sInputFileTrain)
self._dataImpute(self.trainData, self.valData, self.sInputFileVal)
self._dataImpute(self.trainData, self.testData, self.sInputFileTest)
def filterFeatures(self, lDataTypes = []):
"""filters unnecessary features from the dataset
"""
listOrderedSelectedFeatures = [feature for feature in self.trainData.columns if feature in Consts.setSelectedFeatures]
self.trainData = self.trainData[listOrderedSelectedFeatures]
if Consts.DataTypes.VAL in lDataTypes:
self.valData = self.valData[listOrderedSelectedFeatures]
if Consts.DataTypes.TEST in lDataTypes:
self.testData = self.testData[listOrderedSelectedFeatures]
def _changeStringToValues(self, lDataTypes):
"""lDataTypes is a list with following values['test', 'validation'], the list determines if the
_changeStringToValues will happen on the test and validation sets.
After the change, the files are saved with Numeric suffix
"""
self._changeStringToValuesAux(self.trainData, self.sInputFileTrain)
self._changeVoteToNumber(lDataTypes)
# self._fillBoolValues(self.trainData)
# self._fillTrioValues(self.trainData)
# self._fillHotSpot(self.trainData, Consts.listSymbolicColumns)
# # remove previous columns containing strings
# self.trainData = self.trainData.drop(Consts.listNonNumeric, axis=1)
# self.trainData = self.trainData.drop(self.trainData.columns[0], axis=1)
# ElectionsDataPreperation.fixNegativeVals(self.trainData)
# trainPath = self.sInputFileTrain + 'Numeric.csv'
# self.trainData.to_csv(trainPath)
if Consts.DataTypes.TEST in lDataTypes:
self._changeStringToValuesAux(self.testData, self.sInputFileTest)
if Consts.DataTypes.VAL in lDataTypes:
self._changeStringToValuesAux(self.valData, self.sInputFileVal)
def _changeStringToValuesAux(self, data, sInputFilePath):
self._fillBoolValues(data)
self._fillTrioValues(data)
self._fillHotSpot(data, Consts.listSymbolicColumns)
# remove previous columns containing strings
data = data.drop(Consts.listNonNumeric, axis=1)
# data = data.drop(data.columns[0], axis=1)
ElectionsDataPreperation.fixNegativeVals(data)
list_save_to = sInputFilePath.split("/")
data.to_csv(
Consts.FileNames.FILTERED_AND_NUMERIC_NAN.value.format(list_save_to[1], list_save_to[-1].split(".")[0]),
index=False
)
def _refactor_location(self, location: str) -> str:
list_save_to = location.split("/")
return Consts.FileNames.FILTERED_AND_NUMERIC_NAN.value.format(list_save_to[1], list_save_to[-1].split(".")[0])
def _changeVoteToNumber_aux(self, df: pd.DataFrame, location: str):
df[Consts.VOTE_STR] = df[Consts.VOTE_STR].map(Consts.MAP_VOTE_TO_NUMERIC)
header = [Consts.VOTE_STR, Consts.INDEX_COL]
df[header].to_csv(self._refactor_location(location),
header=header,
index=False
)
def _changeVoteToNumber(self, lDataTypes=None):
"""lDataTypes is a list with following values['test', 'validation'], the list determines if the
_changeVoteToNumber will happen on the test and validation sets.
"""
self._changeVoteToNumber_aux(self.trainLabels, self.sInputFileTrainLabel)
if Consts.DataTypes.VAL in lDataTypes:
self._changeVoteToNumber_aux(self.valLabels, self.sInputFileValLabel)
if Consts.DataTypes.TEST in lDataTypes:
self._changeVoteToNumber_aux(self.testLabels, self.sInputFileTestLabel)
def _dataImpute(self, trainData, imputeData, sFileName):
data_with_NaN = imputeData.isnull().any(axis=1)
data_with_NaN = np.where(data_with_NaN)
self.closestFitNanFill(trainData, imputeData, data_with_NaN, sFileName)
def closestFitNanFill(self, trainData, imputeData, data_with_nan, sFileName):
""" finds the closest row to each row that contains NaN and fills the NaNs accordingly
output is saved in a file - "self.sInputFile"
"""
# fill the dict with r value and isNumeric boolean for closest fit function
dist_args_dict = dict()
for feature in trainData.keys():
isNumeric = feature in Consts.setNumericFeatures
d = to_numeric(trainData[feature], errors='coerce')
max_val = d.max(axis=0)
min_val = d.min(axis=0)
dist_args_dict[feature] = (max_val - min_val, isNumeric)
inf = Consts.inf
trainDataArray = trainData.as_matrix()
imputeDataArray = imputeData.as_matrix()
source_null_indexes = None
dest_null_indexes = None
nearestRowDict = dict()
for index in data_with_nan[0]:
source = imputeDataArray[index]
source_null_indexes = np.argwhere(np.isnan(source)).transpose()[0]
nearest_row_rating = inf
for destIndex in range(trainDataArray.shape[0]):
destination = trainDataArray[destIndex]
dest_null_indexes = np.argwhere(np.isnan(destination)).transpose()[0]
if np.intersect1d(source_null_indexes, dest_null_indexes, assume_unique=True).size:
continue
dist = self._distRow(source, destination, dist_args_dict)
# update index for each row with NaN, with nearest neighbour
if dist < nearest_row_rating:
nearest_row_rating = dist
nearestRowDict[index] = destIndex
destination = trainDataArray[nearestRowDict[index]]
source[source_null_indexes] = destination[source_null_indexes]
imputeDataArray[index] = source
# print(source)
print(index.__repr__() + ' ' + nearestRowDict[index].__repr__())
# print(sFileName)
imputeData = pd.DataFrame(imputeDataArray, columns=imputeData.columns.values)
imputeData = imputeData.loc[:, ~imputeData.columns.str.contains('^Unnamed')]
list_save_to = sFileName.split("/")
imputeData.to_csv(
Consts.FileNames.FILTERED_AND_NUMERIC_NONAN.value.format(list_save_to[1], list_save_to[-1].split(".")[0]),
index=False
)
def _fillBoolValues(self, data):
""" replaces Bool columns with 1, 0 and NaN
"""
# no boolean features at all
pass
# data['Looking_at_poles_int'] = data['Looking_at_poles_results'].map({'Yes': 1, 'No': -1, 'NA': np.nan})
# data['Married_int'] = data['Married'].map({'Yes': 1, 'No': -1, 'NA': np.nan})
# data['Gender_int'] = data['Gender'].map({'Male': 1, 'Female': -1, 'NA': np.nan})
# data['Voting_time_int'] = data['Voting_Time'].map({'After_16:00': 1, 'By_16:00': -1, 'NA': np.nan})
# data['Financial_agenda_matters_int'] = data['Financial_agenda_matters'].map({'Yes': 1, 'No': -1, 'NA': np.nan})
def _fillTrioValues(self, data):
data['Will_vote_only_large_party_int'] = data['Will_vote_only_large_party'].map(
{'Yes': 1, 'Maybe': 0, 'No': -1, 'NA': np.nan})
# data['Age_group_int'] = data['Age_group'].map(
# {'Below_30': -1, '30-45': 0, '45_and_up': 1, 'NA': np.nan})
def _fillHotSpot(self, data, featureList):
for feature in featureList:
lFeatures = data[feature].unique()
for category in lFeatures:
if 'nan' == category.__repr__():
continue
dMapping = {x: 1 if x == category else 0 for x in lFeatures}
dMapping['NA'] = np.nan
data[feature + '_' + category] = data[feature].map(dMapping)
def _distFeature(self, xi, yi, r, isContinues):
"""computed distance between features
"""
if np.isnan(xi) or np.isnan(yi):
return 1
if xi == yi:
return 0
if isContinues:
return abs(xi - yi) / r
return 1
def _distRow(self, source, destination, dist_args_dict):
"""computes distance between rows
"""
res = 0
for xi, yi, isNum_r_tuple in zip(source, destination, dist_args_dict.values()):
res += self._distFeature(xi, yi,isNum_r_tuple[0], isNum_r_tuple[1])
return res
def removeAbove95Corr(self):
corr_matrix = self.trainData.corr().abs()
upper = corr_matrix.where(np.triu(np.ones(corr_matrix.shape), k=1).astype(np.bool))
upper.to_csv(self.sInputFileTest+ 'CorrMatrix.csv', index=False)
to_drop = [column for column in upper.columns if any(upper[column] > 0.95)]
self.trainData = self.trainData.drop(to_drop, axis=1)
def sequential_baskward_selection(df: pd.DataFrame, J: callable) -> dict:
base = {feature for feature in df.keys()}
result = {len(base): base}
for i in reversed(range(1, len(base))):
target_feature = max(
[(new_feature, J(result[i + 1].difference({new_feature})))
for new_feature in result[i + 1]],
key=lambda x: x[1])[0]
result[i] = result[i + 1].difference({target_feature})
return result
@staticmethod
def fixNegativeVals(df: pd.DataFrame) -> None:
df[Consts.listFixNegateVals] = np.absolute(df[Consts.listFixNegateVals])
# end class ElectionsDataPreparation
class DataSplit:
""" receives File path, loads data and splits the data in a stratified way
"""
def add_index_col(self, df: pd.DataFrame) -> None:
df[Consts.INDEX_COL] = df.index
def __init__(self, sFilePath):
self.dataset = read_csv(sFilePath, header=0, keep_default_na=True)
self.data = self.dataset.loc[:, self.dataset.columns != 'Vote']
self.add_index_col(self.data)
self.labels = self.dataset.loc[:, self.dataset.columns =='Vote']
self.add_index_col(self.labels)
def saveDataSetsToCsv(self):
""" save splitted data to csv
"""
tDataSets = self.stratifySplit()
for i, dataSet in enumerate(tDataSets):
for j, f in enumerate(Consts.FileSubNames):
dataSet[j].to_csv(Consts.FileNames.RAW_AND_SPLITED.value.format(str(i + 1), f.value, str(i + 1)),
index=False)
def stratifySplit(self):
"""splits the data into three different data sets
"""
X_train_second, X_test_second, y_train_second, y_test_second = train_test_split(self.data, self.labels,
train_size=0.85,
shuffle=True,
random_state=
Consts.listRandomStates[0])
X_train_second, X_val_second, y_train_second, y_val_second = train_test_split(X_train_second,
y_train_second,
train_size=0.8235,
shuffle=True,
random_state=
Consts.listRandomStates[1],
stratify=y_train_second[Consts.VOTE_STR])
X_train_third, X_test_third, y_train_third, y_test_third = train_test_split(self.data, self.labels,
train_size=0.85, shuffle=True,
random_state=
Consts.listRandomStates[2])
X_train_third, X_val_third, y_train_third, y_val_third = train_test_split(X_train_third, y_train_third,
train_size=0.8235, shuffle=True,
random_state=
Consts.listRandomStates[3],
stratify=y_train_third[Consts.VOTE_STR])
return [(X_train_second, X_val_second, X_test_second, y_train_second, y_val_second, y_test_second),
(X_train_third, X_val_third, X_test_third, y_train_third, y_val_third, y_test_third)]
# end method DataSplit