def marginalise_demo2():
"""Generates a GUI to demo marginalisation on a small joint distribution"""
from IO import read_csv
from Data import CompactFactor
from Demos import marginalise_gui
from StringIO import StringIO
cancer = CompactFactor(read_csv(StringIO(cancerdat)))
# create a normal factor
data = cancer['Smoker', 'Cancer', 'Bronchitis']
marginalise_gui(data.normalised())
def florida_demo():
"""Show Florida death penalty data"""
from IO import read_csv
from Data import CompactFactor
import Parameters
from StringIO import StringIO
florida = CompactFactor(read_csv(StringIO(floridadat)))
#create a normal factor
table = florida['Murderer', 'Sentence', 'Victim']
print table
print 'Number of observations is %d' % table.z()
Parameters.precision = 6
print table.normalised()
def cancer_table():
"""Prints out a small contingency table and its normalised version"""
from IO import read_csv
from Data import CompactFactor
import Parameters
from StringIO import StringIO
cancer = CompactFactor(read_csv(StringIO(cancerdat)))
# create a normal factor
table = cancer['Smoker', 'Cancer', 'Bronchitis']
print table
print 'Number of observations is %d' % table.z()
Parameters.precision = 6
print table.normalised()
from random import sample
def random_undersample(dataset, sample_size=-1, sample_percent=-1):
"""Randomply undersamples without replacemnt a dataset."""
# determine sample size
try:
if 0 <= sample_size <= len(dataset):
s_size = int(sample_size)
elif 0 <= sample_percent <= 100:
s_size = int(len(dataset) * sample_percent / 100)
else:
s_size = 0
except:
return []
# return sample
return sample(dataset, s_size)
if __name__ == '__main__':
from IO import read_csv, write_csv, read_header
file_path = '../data/train.csv'
out_file_path = '../data_preprocessed/train_undersampled.csv'
# read dataset
dataset = read_csv(file_path, has_header=True)
header = read_header(file_path)
# undersample
sample = random_undersample(dataset, sample_percent=5)
# write to file
write_csv(out_file_path , sample, header=header)