

In [9]:

    
%reload_ext autoreload
%autoreload 2
%matplotlib inline

import sys
sys.path.append('..')

from helper import linear_regression as lr

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt



In [10]:

    
X, y, Xval, yval, Xtest, ytest = lr.load_data()
# insert the intercept data of every X
X, Xval, Xtest = [np.insert(x.reshape(x.shape[0], 1), 0, np.ones(x.shape[0]), axis=1) for x in (X, Xval, Xtest)]

cost



In [11]:

    
theta = np.ones(X.shape[1])
lr.cost(theta, X, y)









    Out[11]:





303.95152555359761

regularized cost



In [12]:

    
lr.regularized_cost(theta, X, y)









    Out[12]:





303.99319222026429

gradient



In [13]:

    
lr.gradient(theta, X, y)









    Out[13]:





array([ -15.30301567,  598.16741084])

regularized gradient



In [14]:

    
lr.regularized_gradient(theta, X, y)









    Out[14]:





array([ -15.30301567,  598.25074417])

fit the data

regularization term $\lambda=0$



In [15]:

    
theta = np.ones(X.shape[0])

final_theta = lr.linear_regression_np(X, y, l=0).get('x')



In [16]:

    
b = final_theta[0] # intercept
m = final_theta[1] # slope

plt.scatter(X[:,1], y, label="Training data")
plt.plot(X[:, 1], X[:, 1]*m + b, label="Prediction")
plt.legend(loc=2)









    Out[16]:





<matplotlib.legend.Legend at 0x11a196fd0>



In [ ]: