-
Notifications
You must be signed in to change notification settings - Fork 0
/
Assignment 5 Q1.py
144 lines (115 loc) · 3.65 KB
/
Assignment 5 Q1.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
# -*- coding: utf-8 -*-
"""
Created on Sat Nov 5 18:31:58 2022
@author: SANJUSHA
"""
import numpy as np
import pandas as pd
df = pd.read_csv("50_Startups.csv")
df.head()
df1=df.rename({'R&D Spend':'RDS','Administration':'ADMS','Marketing Spend':'MKTS'},axis=1)
df1
df1.corr()
# Profit and RDS has the highest correlation of 0.9729 following MKTS and ADMS
# Splitting the variables
Y = df1[["Profit"]]
# X = df1[["RDS"]] # Model 1
# X = df1[["RDS","ADMS"]] # Model 2
X = df1[["RDS","ADMS","MKTS"]] # Model 3
import matplotlib.pyplot as plt
plt.scatter(x = X["RDS"],y = Y, color= 'black')
plt.show()
# Here, the plot shows strong positive correlation
import matplotlib.pyplot as plt
plt.scatter(x = X["ADMS"],y = Y, color= 'black')
plt.show()
# Here, the plot shows low correlation
import matplotlib.pyplot as plt
plt.scatter(x = X["MKTS"],y = Y, color= 'black')
plt.show()
# Here, the plot shows moderate positive correlation
# Boxplot
df1.boxplot("RDS",vert=False)
df1.boxplot("ADMS",vert=False)
df1.boxplot("MKTS",vert=False)
# There are no outliers
# Train and Test
from sklearn.model_selection import train_test_split
X_train,X_test,Y_train,Y_test=train_test_split(X,Y,test_size=0.2)
from sklearn.linear_model import LinearRegression
LR = LinearRegression()
LR.fit(X_train,Y_train)
# Predictions
Y_predtrain=LR.predict(X_train)
Y_predtest=LR.predict(X_test)
from sklearn.metrics import mean_squared_error, r2_score
mse1 = mean_squared_error(Y_train,Y_predtrain)
mse2 = mean_squared_error(Y_test,Y_predtest)
rmse1 = np.sqrt(mse1).round(2)
rmse2 = np.sqrt(mse2).round(2)
r2_score1 = r2_score(Y_train,Y_predtrain)
r2_score2 = r2_score(Y_test,Y_predtest)
# Model Validation - Multicollinearity, KFold
# pip install statsmodels
import statsmodels.api as sma
X_new = sma.add_constant(X)
lm = sma.OLS(Y,X_new).fit()
lm.summary()
# p value of ADMS, MKTS is >0.05. Therefore it has multicollinearity issues
from sklearn.model_selection import KFold, cross_val_score
k=13
k_fold=KFold(n_splits=k, random_state=None)
cv_scores=cross_val_score(LR, X_train, Y_train, cv=k_fold)
mean_acc_score=sum(cv_scores)/len(cv_scores)
# Model Deletion - Cooks Distance
# Suppress scientific notation
import numpy as np
np.set_printoptions(suppress=True)
# Create instance of influence
influence = lm.get_influence()
# Obtain Cook's distance for each observation
cooks = influence.cooks_distance
# Display Cook's distances
print(cooks)
import matplotlib.pyplot as plt
plt.scatter(df1.RDS, cooks[0])
plt.xlabel('X')
plt.ylabel('Cooks Distance')
plt.show()
import matplotlib.pyplot as plt
plt.scatter(df1.ADMS, cooks[0])
plt.xlabel('X')
plt.ylabel('Cooks Distance')
plt.show()
import matplotlib.pyplot as plt
plt.scatter(df1.MKTS, cooks[0])
plt.xlabel('X')
plt.ylabel('Cooks Distance')
plt.show()
# MODEL 1
# MSE1 = 81445821.5764
# MSE2 = 80510457.0198
# RMSE1 = 9024.73
# RMSE2 = 8272.76
# r2_score1 = 0.96321
# r2_score2 = 0.94616
# Mean_accuracy = 95%
# MODEL 2
# MSE1 = 87479410.1205
# MSE2 = 68761805.0140
# RMSE1 = 9353.04
# RMSE2 = 8592.27
# r2_score1 = 0.94505
# r2_score2 = 0.93779
# Mean_accuracy = 94%
# MODEL 3
# MSE1 = 88540951.2358
# MSE2 = 75875986.8478
# RMSE1 = 9409.62
# RMSE2 = 8710.68
# r2_score1 = 0.94937
# r2_score2 = 0.89819
# Mean_accuracy = 91%
# Inference : Here Model 1 where Y=df1[["Profit"]] and X=df1[["RDS"]] is selected for profit of
# 50_startups data, since its r2 for train and test are 0.96321 and 0.94616, mean_accuracy = 95%,
# mse and rmse and lower than the other models and for less expense and more profit the first model is the best.