import numpy as np
import pandas as pd

def calculate_w(x,y):
   col_ones = np.ones((11800,1))
   x_modified = np.hstack((x,col_ones))

   x_pseudo_inversed = np.linalg.pinv(x_modified)
   w = np.matmul(x_pseudo_inversed, y)
   return w


def predict(w,new_x):
    predicted_value = np.matmul(w,new_x)
    if predicted_value > 0:
        print ("c1")
    else:
        print ("c2")



def main():
    y = []
    x = []
    dataset = pd.read_csv('mnist_train.csv')
    dataset = dataset.as_matrix()

    for i in range(len(dataset)):
        if dataset[i][0] == 8 or dataset[i][0] == 9:
            y.append(dataset[i][0])
            x.append(dataset[i][1:].tolist())

    x = np.array(x)
    y = np.array(y)
    w = calculate_w(x,y)