-
Notifications
You must be signed in to change notification settings - Fork 0
/
predict.py
179 lines (148 loc) · 9.36 KB
/
predict.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
# -*- coding: utf-8 -*-
"""
Created on Sun Nov 3 19:56:06 2019
@author: Akash Verma
"""
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn import metrics
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import OneHotEncoder
import seaborn as sns
import category_encoders as ce
from sklearn.preprocessing import LabelBinarizer
#import xgboost as xgb
from sklearn.ensemble import RandomForestRegressor
#To load Input data
dataset_test = pd.read_csv('tcd-ml-1920-group-income-train.csv')
dataset_pred = pd.read_csv('tcd-ml-1920-group-income-test.csv')
store_data = dataset_pred.filter(['Instance'], axis=1)
dataset_test.columns = dataset_test.columns.str.strip().str.replace(' ', '_').str.replace('(', '').str.replace(')', '')
dataset_pred.columns = dataset_pred.columns.str.strip().str.replace(' ', '_').str.replace('(', '').str.replace(')', '')
#print(dataset_test.shape)
dataset_test.rename(columns={'Work_Experience_in_Current_Job_[years]': 'Work_Experience_in_Current_Job'}, inplace=True)
dataset_pred.rename(columns={'Work_Experience_in_Current_Job_[years]': 'Work_Experience_in_Current_Job'}, inplace=True)
#print(dataset_test.columns)
#Remove Instance Column
dataset_test = dataset_test.drop('Instance', 1)
dataset_pred = dataset_pred.drop('Instance', 1)
dataset_test.drop(dataset_test.loc[dataset_test['Total_Yearly_Income_[EUR]']==2548791].index, inplace=True)
#Check if year is null
#print('NUll records in year: ',dataset_test['Year_of_Record'].isnull().sum())
#Change null to rand value between std dev around mean. Alternate could be panda interpolate
Column_Name_avg = dataset_test['Year_of_Record'].mean()
Column_Name_std = dataset_test['Year_of_Record'].std()
Column_Name_null_count = dataset_test['Year_of_Record'].isnull().sum()
Column_Name_null_random_list = np.random.randint(Column_Name_avg - Column_Name_std, Column_Name_avg + Column_Name_std, size=Column_Name_null_count)
dataset_test['Year_of_Record'][np.isnan(dataset_test['Year_of_Record'])] = Column_Name_null_random_list
dataset_test['Year_of_Record'] = dataset_test['Year_of_Record'].astype(int)
Column_Name_avg = dataset_pred['Year_of_Record'].mean()
Column_Name_std = dataset_pred['Year_of_Record'].std()
Column_Name_null_count = dataset_pred['Year_of_Record'].isnull().sum()
Column_Name_null_random_list = np.random.randint(Column_Name_avg - Column_Name_std, Column_Name_avg + Column_Name_std, size=Column_Name_null_count)
dataset_pred['Year_of_Record'][np.isnan(dataset_pred['Year_of_Record'])] = Column_Name_null_random_list
dataset_pred['Year_of_Record'] = dataset_pred['Year_of_Record'].astype(int)
#Filling Gender Nan and other values
dataset_test.Gender = dataset_test.Gender.replace("f", "female")
dataset_test["Gender"].fillna("unknown", inplace = True)
dataset_test.Gender = dataset_test.Gender.replace("0", "unknown")
dataset_pred.Gender = dataset_pred.Gender.replace("f", "female")
dataset_pred["Gender"].fillna("unknown", inplace = True)
dataset_pred.Gender = dataset_pred.Gender.replace("0", "unknown")
#Filling University Degree Nan Values
dataset_test.University_Degree = dataset_test.University_Degree.replace("0", "No")
dataset_test["University_Degree"].fillna("No", inplace = True)
dataset_pred.University_Degree = dataset_pred.University_Degree.replace("0", "No")
dataset_pred["University_Degree"].fillna("No", inplace = True)
#Filling Haircolor Nan Values
dataset_test.Hair_Color = dataset_test.Hair_Color.replace("0", "Unknown")
dataset_test["Hair_Color"].fillna("Unknown", inplace = True)
dataset_test = dataset_test.drop('Hair_Color', 1)
dataset_pred.Hair_Color = dataset_pred.Hair_Color.replace("0", "Unknown")
dataset_pred["Hair_Color"].fillna("Unknown", inplace = True)
dataset_pred = dataset_pred.drop('Hair_Color', 1)
#Filling Profession Nan Values
dataset_test["Profession"].fillna("Unknown", inplace = True)
dataset_pred["Profession"].fillna("Unknown", inplace = True)
#Filling Housing_Situation Corrupt Values
dataset_test["Housing_Situation"] = dataset_test["Housing_Situation"].replace(0,'Unknown')
dataset_test["Housing_Situation"] = dataset_test["Housing_Situation"].replace('0','Unknown')
dataset_test["Housing_Situation"] = dataset_test["Housing_Situation"].replace('nA','Unknown')
dataset_pred["Housing_Situation"] = dataset_pred["Housing_Situation"].replace(0,'Unknown')
dataset_pred["Housing_Situation"] = dataset_pred["Housing_Situation"].replace('0','Unknown')
dataset_pred["Housing_Situation"] = dataset_pred["Housing_Situation"].replace('nA','Unknown')
dataset_test['Satisfation_with_employer'] = dataset_test['Satisfation_with_employer'].fillna(method='ffill')
dataset_pred['Satisfation_with_employer'] = dataset_pred['Satisfation_with_employer'].fillna(method='ffill')
#Convert work Experience in current job to float and replace #NUM! to mean of their Age
dataset_test.Work_Experience_in_Current_Job = dataset_test.Work_Experience_in_Current_Job.replace("#NUM!", None)
dataset_test['Work_Experience_in_Current_Job'] = dataset_test['Work_Experience_in_Current_Job'].astype(float)
dataset_test.value = dataset_test.groupby('Age')['Work_Experience_in_Current_Job'].apply(lambda x: x.fillna(x.median()))
dataset_test.value = dataset_test.Work_Experience_in_Current_Job.fillna(dataset_test.Work_Experience_in_Current_Job.median())
dataset_pred.Work_Experience_in_Current_Job = dataset_pred.Work_Experience_in_Current_Job.replace("#NUM!", None)
dataset_pred['Work_Experience_in_Current_Job'] = dataset_pred['Work_Experience_in_Current_Job'].astype(float)
dataset_pred.value = dataset_pred.groupby('Age')['Work_Experience_in_Current_Job'].apply(lambda x: x.fillna(x.median()))
dataset_pred.value = dataset_pred.Work_Experience_in_Current_Job.fillna(dataset_pred.Work_Experience_in_Current_Job.median())
#Remove EUR in Yearly_Income to make it numerical value
dataset_test['Yearly_Income_in_addition_to_Salary_e.g._Rental_Income'] = dataset_test['Yearly_Income_in_addition_to_Salary_e.g._Rental_Income'].replace('EUR', '', regex=True).astype(float)
dataset_pred['Yearly_Income_in_addition_to_Salary_e.g._Rental_Income'] = dataset_pred['Yearly_Income_in_addition_to_Salary_e.g._Rental_Income'].replace('EUR', '', regex=True).astype(float)
#dataset_test['Salary_by_Year'] = dataset_test.groupby('Year_of_Record')['Total_Yearly_Income_[EUR]'].transform('median')
#dataset_test['Salary_by_Profession'] = dataset_test.groupby('Profession')['Total_Yearly_Income_[EUR]'].transform('mean')
#Label encoding the categorical columns
for col in dataset_test.dtypes[dataset_test.dtypes == 'object'].index.tolist():
feat_le = LabelEncoder()
train_list = dataset_test[col].unique()
dataset_test.loc[1201,col] = 'other'
test_list = dataset_pred[col].unique()
test_replace = list(set(test_list) - set(train_list))
dataset_pred[col] = dataset_pred[col].replace(test_replace, 'other')
feat_le.fit(dataset_test[col].unique().astype(str))
dataset_test[col] = feat_le.transform(dataset_test[col].astype(str))
dataset_pred[col] = feat_le.transform(dataset_pred[col].astype(str))
#this function takes a string column name and returns a list
#containing indices of dataframe that have outliers in that column
#Refer: https://towardsdatascience.com/5-ways-to-detect-outliers-that-every-data-scientist-should-know-python-code-70a54335a623
def OutlierByColumn(colname):
rows = dataset_test.shape[0]
col_std = np.std(dataset_test[colname])
col_mean = np.mean(dataset_test[colname])
anomaly_cut_off = col_std * 3
lower_limit = col_mean - anomaly_cut_off
upper_limit = col_mean + anomaly_cut_off
anomalies_indices = []
for i in range(rows):
ele = dataset_test[colname][i]
if ele > upper_limit or ele < lower_limit:
anomalies_indices.append(i)
return anomalies_indices
arr1 = OutlierByColumn('Work_Experience_in_Current_Job')
arr2 = OutlierByColumn('Age')
#arr3 = OutlierByColumn('Body_Height_[cm]')
arr4 = OutlierByColumn('Yearly_Income_in_addition_to_Salary_e.g._Rental_Income')
arr5 = OutlierByColumn('Size_of_City')
#Union of all lists
union_index = np.union1d(arr1,arr2)
#union_index = np.union1d(union_index,arr3)
union_index = np.union1d(union_index,arr4)
union_index = np.union1d(union_index,arr5)
len(union_index)
dataset_test = dataset_test.drop(union_index)
y = dataset_test['Total_Yearly_Income_[EUR]'].values
dataset_test.drop('Total_Yearly_Income_[EUR]', axis=1, inplace=True)
X = dataset_test[dataset_test.columns].values
#split 80% of the data to the training set while 20% of the data to test set.
#X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=0)
X_train = X
y_train = y
dataset_pred.drop('Total_Yearly_Income_[EUR]', axis=1, inplace=True)
X_test = dataset_pred[dataset_pred.columns].values
regressor= RandomForestRegressor(n_estimators=100,random_state=42)
regressor.fit(X_train, y_train)
y_pred = regressor.predict(X_test)
#print('Mean Absolute Error:', metrics.mean_absolute_error(y_test, y_pred))
#print('Mean Squared Error:', metrics.mean_squared_error(y_test, y_pred))
#print('Root Mean Squared Error:', np.sqrt(metrics.mean_squared_error(y_test, y_pred)))
store_data['Total Yearly Income [EUR]'] = y_pred
store_data.to_csv('output.csv', sep=',')