split

#! /usr/bin/python3

from random import randint

f = open("leukemia.dat", "r")

lines = sum(1 for line in open('leukemia.dat'))
# get 20% of toal lines
tot = int((int(lines)*20)/100)
rand_lines = []

for _ in range(tot):
    #generate random value
    n =  randint(1, lines)
    while n in rand_lines:
        n =  randint(1, lines)
    # add to the list
    rand_lines.append(n)

test = open("test.dat","w+")
train = open("train.dat","w+")

i=0
for x in f:
    if i == 0:
        test.write(x)
        train.write(x)
    elif i in rand_lines:
        test.write(x)
    else:
        train.write(x)

    i += 1

print("Done!")
f.close()