-
Notifications
You must be signed in to change notification settings - Fork 0
/
regression.py
72 lines (58 loc) · 2.41 KB
/
regression.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
import pandas as pd
import numpy as np
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, r2_score
from sklearn.datasets import load_boston
class CustomLinearRegression:
def __init__(self, fit_intercept=True):
self.fit_intercept = fit_intercept
self.coefficient = np.array([])
self.intercept = 0.0
self.r2 = ...
self.rmse_ = ...
self.features = np.matrix
def fit(self, X, y):
if self.fit_intercept:
X['ones'] = [1 for _ in range(X.shape[0])]
self.features = np.matrix(X.values)
X = self.features
y = np.matrix(y.values)
beta = np.linalg.inv(X.T @ X) @ X.T @ y
n = beta.shape[0]
self.intercept = beta[n-1, 0]
self.coefficient = np.array(beta[:n-1])
else:
self.features = np.matrix(X.values)
X = self.features
y = np.matrix(y.values)
self.coefficient = np.linalg.inv(X.T @ X) @ X.T @ y
def predict(self, X):
return np.matrix(X.values) @ np.matrix(self.coefficient) + self.intercept
def r2_score(self, y, yhat):
self.r2 = (1 - (np.sum(np.subtract(y, yhat) ** 2) / np.sum(np.subtract(y, y.mean()) ** 2))).array[0]
def rmse(self, y, yhat):
self.rmse_ = ((np.sum(np.subtract(y, yhat) ** 2) / len(y)) ** 0.5).array[0]
def main():
data = load_boston()
X, y = data.data, data.target
X_train = X[:-100, :]
y_train = y[:-100]
X_test = X[-100:, :]
y_test = y[-100:]
customModel = CustomLinearRegression(fit_intercept=True)
customModel.fit(pd.DataFrame(X_train), pd.DataFrame(y_train))
y_pred1 = customModel.predict(pd.DataFrame(X_test))
customModel.r2_score(pd.DataFrame(y_test), y_pred1)
customModel.rmse(pd.DataFrame(y_test), y_pred1)
skModel = LinearRegression(fit_intercept=True)
skModel.fit(X_train, y_train)
y_pred2 = skModel.predict(X_test)
r2_sk = r2_score(y_test, y_pred2)
rmse_sk = mean_squared_error(y_test, y_pred2) ** 0.5
output = {'Intercept': skModel.intercept_ - customModel.intercept,
'Coefficient': skModel.coef_ - customModel.coefficient.T,
'R2': r2_sk - customModel.r2,
'RMSE': rmse_sk - customModel.rmse_}
print(output)
if __name__ == '__main__':
main()