def average_error(design_columns, target_columns, N):
error = 0
for i in range(0,N):
splitter.create_training_and_test_sets('data/bodyfatdata.txt', training_ratio = 0.8)
model = maximum_likelihood_model('data/bodyfat_training.txt',
design_columns, target_columns)
error += root_mean_square_error('data/bodyfat_test.txt',
design_columns, target_columns, model)
return error / float(N)
#!/usr/bin/python
import classification as cla
import regression as reg
import split_data as spl
import numpy as np
if (raw_input("II.1 Split the bodyfat dataset into training and test sets? (y/n):") == "y"):
spl.create_training_and_test_sets('data/bodyfatdata.txt')
print "... Saved datasets as data/bodyfat_training.txt and data/bodyfat_test.txt"
print
if (raw_input("II1.1 Compute the RMS for the maximum likelihood solution? (y/n):") == "y"):
ml_model1 = reg.maximum_likelihood_model('data/bodyfat_training.txt',
[3, 6, 7, 8], [1])
ml_model2 = reg.maximum_likelihood_model('data/bodyfat_training.txt', [7], [1])
ml_error1 = reg.root_mean_square_error('data/bodyfat_test.txt',
[3, 6, 7, 8], [1], ml_model1)
ml_error2 = reg.root_mean_square_error('data/bodyfat_test.txt', [7], [1], ml_model2)
print "The RMS error of the test set for selection 1:"
print ml_error1
print "The RMS error of the test set for selection 2:"
print ml_error2
print
if (raw_input("II1.2 Compute the RMS for the maximum a posteriori solution? (y/n):") == "y"):
reg.plot_alpha_error('data/bodyfat_training.txt','data/bodyfat_test.txt')
print "Saved plots as images/rms_selection1.png and images/rms_selection2.png"
print
