-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdata_preprocess.py
174 lines (141 loc) · 8.99 KB
/
data_preprocess.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
import pandas as pd
import numpy as np
import seaborn as sns
from sklearn.preprocessing import LabelEncoder
from sklearn.decomposition import PCA
from fancyimpute import KNN, IterativeImputer
import os
os.chdir('/Users/bratislavpetkovic/Desktop/My_Kick_Ass_Portfolio/Housing-Price-Regression/')
test_df = pd.read_csv('house-prices-advanced-regression-techniques/test.csv')
original_train_df = pd.read_csv('house-prices-advanced-regression-techniques/train.csv')
print(original_train_df.info())
print(test_df.info())
continuous_columns = ['LotFrontage', 'LotArea', 'OverallQual', 'OverallCond', 'YearBuilt', 'YearRemodAdd', 'MasVnrArea', 'BsmtFinSF1', 'BsmtFinSF2',
'BsmtUnfSF' , '1stFlrSF', '2ndFlrSF', 'LowQualFinSF', 'GrLivArea', 'BsmtFullBath', 'BsmtHalfBath', 'FullBath', 'HalfBath',
'TotRmsAbvGrd', 'Fireplaces', 'GarageYrBlt', 'GarageCars', 'WoodDeckSF', 'OpenPorchSF', 'EnclosedPorch', '3SsnPorch', 'ScreenPorch', 'PoolArea',
'MiscVal', 'MoSold', 'YrSold', 'MSSubClass', 'GarageArea', 'SalePrice', 'KitchenAbvGr', 'BedroomAbvGr', 'TotalBsmtSF' ]
#_______________________________________WRANGLING + IMPUTATION + DIMENSIONALITY REDUCTION_______________________________________
def wrangle_data(data, imputation_method='KNN', dim_reduction = True, is_test=False):
cols_2_remove = [col for col in data.columns if data[col].isna().sum()/data.shape[0] > 0.6 ] # rid columsn where more than 60% data is missing
cols_2_impute = [col for col in data.columns if 0 < data[col].isna().sum()/data.shape[0] < 0.6 ] # impute other columsn with missing data
columns_2_keep = list(set(data.columns) - set(cols_2_remove))
data = data[columns_2_keep]
# using trees so refrain from one-hot encoding
one_hot_cols = [ 'Street' ]
nominal_cols = [ 'MSZoning', 'Neighborhood', 'Condition1', 'Condition2', 'HouseStyle', 'RoofStyle', 'LotConfig', 'BldgType', 'RoofMatl', 'Exterior1st',
'Exterior2nd', 'MasVnrType', 'Foundation', 'Heating', 'Electrical', 'GarageType', 'SaleType', 'SaleCondition' ]
ordinal_cols = [ 'LotShape', 'LandContour', 'Utilities', 'LandSlope', 'ExterQual', 'ExterCond', 'BsmtQual', 'BsmtCond', 'BsmtExposure', 'BsmtFinType1',
'BsmtFinType2', 'HeatingQC', 'CentralAir', 'KitchenQual', 'Functional', 'FireplaceQu', 'GarageFinish', 'GarageQual', 'GarageCond', 'PavedDrive']
remaing_cols = continuous_columns + ['Id']
if(is_test):
remaing_cols.remove('SalePrice')
print('did we forget any columns? --> ', len(set(list(one_hot_cols + nominal_cols + ordinal_cols + remaing_cols) )) != len(data.columns) )
print('columns forgotten', set(data.columns) - set(list(one_hot_cols + nominal_cols + ordinal_cols + remaing_cols) ))
print('columns repeated', set(list(one_hot_cols + nominal_cols + ordinal_cols + remaing_cols) ) - set(data.columns) )
# 1 Hot Encoder
one_hot_df_copy = data[one_hot_cols+['Id']]
prefixes = ['' for x in one_hot_cols]
one_hot_df = pd.get_dummies(one_hot_df_copy, prefix= prefixes, columns = one_hot_cols, drop_first=True)
print("# of New Columns Added: ", one_hot_df.shape[1] - one_hot_df_copy.shape[1])
print("Columns Added: ", set(one_hot_df.columns) - set(one_hot_df.columns))
print("Columns Lost: ", set(one_hot_df.columns) - set(one_hot_df.columns))
# LabelEncoding
label_df = data[nominal_cols]
label_df = label_df.apply(lambda series: pd.Series(LabelEncoder().fit_transform(series[series.notnull()]),index=series[series.notnull()].index))
label_df["Id"] = data['Id'].copy()
# print(label_df.info())
print(label_df.Id.max())
print(label_df.Id.min())
print(data.Id.max())
print(data.Id.min())
# OrdinalEncoding
ordinal_df = data[ordinal_cols+remaing_cols]
lot_shape_dic = {'IR3':0, 'IR2':1, 'IR1':2, 'Reg':3}
land_cont_dic = {'Low':0, 'HLS':1, 'Bnk':2, 'Lvl':3}
utilities_dic = {'ELO':0, 'NoSeWa':1, 'NoSewr':2, 'AllPub':3}
land_slope_dic = {'Sev':0, 'Mod':1, 'Gtl':2}
quality_cond_dic = {np.nan:-1, 'NA':-1, 'Po':0, 'Fa':1, 'TA':2, 'Gd':3, 'Ex':4}
bsmnt_exp_dict = {np.nan:-1, 'No':0, 'Mn':1, 'Av':2, 'Gd':3}
bsmnt_fin = {np.nan:-1, 'Unf':0, 'LwQ':1, 'BLQ':2, 'Rec':3, 'ALQ':4, 'GLQ':5}
y_n_dic = {'N':0, 'Y':1}
functional_dic = {'Sal':0, 'Sev':1, 'Maj2':2, 'Maj1':3, 'Mod':4, 'Min2':5, 'Min1':6, 'Typ':7 }
gar_fin_dict = {np.nan:0,'NA':0, 'Unf':1, 'RFn':2, 'Fin':3}
pav_dr_dict = {'N':0, 'P':1, 'Y':2}
ordinal_df['LotShape'].replace(lot_shape_dic, inplace=True)
ordinal_df['LandContour'].replace(land_cont_dic, inplace=True)
ordinal_df['Utilities'].replace(utilities_dic, inplace=True)
ordinal_df['LandSlope'].replace(land_slope_dic, inplace=True)
ordinal_df['ExterQual'].replace(quality_cond_dic, inplace=True)
ordinal_df['ExterCond'].replace(quality_cond_dic, inplace=True)
ordinal_df['BsmtQual'].replace(quality_cond_dic, inplace=True) #
ordinal_df['BsmtCond'].replace(quality_cond_dic, inplace=True) #
ordinal_df['BsmtExposure'].replace(bsmnt_exp_dict, inplace=True)
ordinal_df['BsmtFinType1'].replace(bsmnt_fin, inplace=True)
ordinal_df['BsmtFinType2'].replace(bsmnt_fin, inplace=True)
ordinal_df['HeatingQC'].replace(quality_cond_dic, inplace=True)
ordinal_df['CentralAir'].replace(y_n_dic, inplace=True)
ordinal_df['KitchenQual'].replace(quality_cond_dic, inplace=True)
ordinal_df['Functional'].replace(functional_dic, inplace=True)
ordinal_df['FireplaceQu'].replace(quality_cond_dic, inplace=True) #
ordinal_df['GarageFinish'].replace(gar_fin_dict, inplace=True)
ordinal_df['GarageQual'].replace(quality_cond_dic, inplace=True) #
ordinal_df['GarageCond'].replace(quality_cond_dic, inplace=True) #
ordinal_df['PavedDrive'].replace(pav_dr_dict, inplace=True)
# Merge data back together
one_hot_df_cols = set(one_hot_df.columns)
ordinal_df_cols = set(ordinal_df.columns)
label_df_cols = set(label_df.columns)
intersect = one_hot_df_cols.intersection(ordinal_df_cols,label_df_cols )
print("One Hot Encoded: ", one_hot_df_cols)
print("Ordinal Encoded: ", ordinal_df_cols)
print("Label Encoded: ", label_df_cols)
data = one_hot_df.merge(ordinal_df, how="outer", on="Id")
data = data.merge(label_df, how="outer", on="Id")
print(data.info())
data = data.drop(columns=['Id'])
print(data.shape)
#_______________________________________IMPUTATION_______________________________________
if(imputation_method!='KNN'):
# MICE IMPUTATION
mice_imputer = IterativeImputer()
data.iloc[:, :] = mice_imputer.fit_transform(data)
else:
# KNN IMPUTATION
knn_imputer = KNN()
data.iloc[:, :] = knn_imputer.fit_transform(data)
print(data.shape)
#_______________________________________Dim Reduction_______________________________________
if(dim_reduction):
# ANALYSIS OF VARIABLES AND THEIR CORRELATION
sns.set(rc={'figure.figsize':(40,40)})
cmap=sns.diverging_palette(h_neg=10, h_pos=240, as_cmap=True)
mask = np.triu(np.ones_like(data.corr().abs(), dtype=bool))
dataset_corr = sns.heatmap(data.corr().abs(), mask=mask, center=0, cmap=cmap, linewidths=1, annot=True, fmt='.2f')
dataset_corr.get_figure().savefig('check_me_out/column_correlation.png')
drop_candidates = ['Exterior2nd', 'GarageCond', 'GarageCars', 'FireplaceQu']
data = data.drop(columns=drop_candidates)
return data
def run_pca(data, continuous_columns):
#_______________________________________PCA_______________________________________
pca_subset_df = data[continuous_columns]
# NORMALIZE Data to improve PCA perform.
df_std = (pca_subset_df - pca_subset_df.mean())/ pca_subset_df.std()
pca = PCA(n_components = df_std.shape[1]) # 0.9. or medical_cont_df.shape[1]
pca.fit_transform(df_std)
columns_gen_2 = ['PC'+ str(x) for x in range(1,len(df_std.columns)+1)]
print(pca.explained_variance_ratio_.cumsum())
loadings = pd.DataFrame( pca.components_.T, columns = columns_gen_2, index = pca_subset_df.columns )
print(loadings)
return loadings
#_______________________________________OUTPUT_______________________________________
train_knn_df = wrangle_data(original_train_df, dim_reduction=False)
pca = run_pca(train_knn_df, continuous_columns)
train_knn_df.to_csv('wrangled_data/train_knn_df.csv', index=False, )
train_knn_reduced_df = wrangle_data(original_train_df)
train_knn_reduced_df.to_csv('wrangled_data/train_knn_reduced_df.csv', index=False,)
train_mice_df = wrangle_data(original_train_df, dim_reduction=False, imputation_method='MICE')
train_mice_df.to_csv('wrangled_data/train_mice_df.csv', index=False,)
train_mice_reduced_df = wrangle_data(original_train_df, imputation_method='MICE')
train_mice_df.to_csv('wrangled_data/train_mice_reduced_df.csv', index=False,)
test_prepped_df = wrangle_data(test_df, is_test=True, dim_reduction=False)
test_prepped_df.to_csv('wrangled_data/test_prepped_df.csv', index=False)