notebook.community

Edit and run



In [1]:

    
%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 [2]:

    
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)]



In [3]:

    
training_cost, cv_cost = [], []

use the subset of training set to fit the model
no regularization when you compute the training cost and CV cost
remember to use the same subset of training set to compute training cost



In [4]:

    
m = X.shape[0]
for i in range(1, m+1):
#     print('i={}'.format(i))
    res = lr.linear_regression_np(X[:i, :], y[:i], l=0)
    
    tc = lr.regularized_cost(res.x, X[:i, :], y[:i], l=0)
    cv = lr.regularized_cost(res.x, Xval, yval, l=0)
#     print('tc={}, cv={}'.format(tc, cv))
    
    training_cost.append(tc)
    cv_cost.append(cv)



In [5]:

    
plt.plot(np.arange(1, m+1), training_cost, label='training cost')
plt.plot(np.arange(1, m+1), cv_cost, label='cv cost')
plt.legend(loc=1)









    Out[5]:





<matplotlib.legend.Legend at 0x117fba7b8>

the model can't even fit traning set well, under fitting



In [ ]: