Linear Regression

lin_reg.py

import sys
import numpy
import array

factors, numTraining = map(int, sys.stdin.readline().split(' '))

trainSets = []
for i in range(numTraining):
	trainSets.append(map(float, sys.stdin.readline().split(' ')))

rows = numpy.array(trainSets)
rows_indxs = [[i] for i in range(numTraining)]
factor_indxs = [i for i in range(factors)]
# print rows_indxs
# print factor_indxs
y = rows[rows_indxs,[factors]]
x = rows[rows_indxs,factor_indxs]
zeros = numpy.zeros((numTraining, 1))
# training = numpy.hstack((zeros, x))
training = numpy.hstack((zeros, rows))


weights = [1 for i in range(factors + 1)]

def d_training_item(trainObs, weights):
	predicted_y = sum([ weights[i] * trainObs[i] for i in range(factors)])
	return abs(predicted_y - trainObs[factors]) / trainObs[factors]

def calc_d(weights, training):
	return sum([d_training_item(trainObs, weights) for trainObs in training])

print calc_d(weights, training)