run_mimify_GAN.py

from CCIT import *
import numpy as np
import pandas as pd
from src.MIMIFY_GAN import *
from src.MIMIFY_REG import *

import argparse


def get_args():
    parser = argparse.ArgumentParser(
        description='runs CI test from a folder')
    # Add arguments
    parser.add_argument(
        '-fp', '--folder_path', type=str, help='Data Folder Path', required=True)
    parser.add_argument(
        '-dx', '--dimX', type=int, help='dimension of X', required=True)
    parser.add_argument(
        '-dy', '--dimY', type=int, help='dimension of Y', required=True)
    parser.add_argument(
        '-dz', '--dimZ', type=int, help='dimension of X', required=True)
    parser.add_argument(
        '-dn', '--dimN', type=int, help='dimension of noise', required=False)
    parser.add_argument(
        '-nf', '--numfile', type=int, help='number of files', required=True)
    parser.add_argument(
        '-nt', '--nthread', type=int, help='number of threads', required=False)
    parser.add_argument(
        '-me', '--maxepoch', type=int, help='max. number of epochs to train regressor', required=False)
    parser.add_argument(
        '-bs', '--bsize', type=int, help='batch size', required=False)
    parser.add_argument(
        '-rf', '--rfile', type=str, help='result file name', required=True)
    parser.add_argument(
        '-nm', '--normalized', type=int, help='normalize data-set?', required=True)
    parser.add_argument(
        '-dm', '--dim', type=int, help='hidden size of regressor', required=False)
    parser.add_argument(
        '-dc', '--deepclassifier', type=int, help='1 if deep classifier, 0 if xgb', required=True)
    parser.add_argument(
        '-nh', '--nhid', type=int, help='number of neurons in hidden layer of deep classifier', required=False)
    parser.add_argument(
        '-nl', '--nlayers', type=int, help='number of layers in deep classifier', required=False)
    parser.add_argument(
        '-dr', '--dropout', type=float, help='dropout in deep classifier', required=False)

    # Array for all arguments passed to script
    args = parser.parse_args()
    folder_path = args.folder_path
    dx = args.dimX
    dy = args.dimY
    dz = args.dimZ
    numfile = args.numfile
    if args.nthread:
        nthread = args.nthread
    else:
        nthread = 16

    if args.dimN:
        dimN = args.dimN 
    else:
        dimN = 30

    if args.maxepoch:
        maxepoch = args.maxepoch
    else:
        maxepoch = 50
    if args.bsize:
        bsize = args.bsize
    else:
        bsize = 50
    rfile = args.rfile
    normalized = args.normalized
    dc = args.deepclassifier
    if args.nhid:
        nh = args.nhid
    else:
        nh = 20
    if args.nlayers:
        nl = args.nlayers
    else:
        nl = 5
    if args.dropout:
        dr = args.dropout
    else:
        dr = 0.2


    return folder_path,dx,dy,dz,numfile,nthread,dimN,maxepoch,bsize,rfile,normalized,dc,nh,nl,dr




if __name__ == "__main__":
    folder_path,dx,dy,dz,numfile,nthread,dim_N,maxepoch,bsize,rfile,normalized,dc,nh,nl,dr = get_args()
    if normalized == 1:
        temp = True
    else:
        temp = False
    if dc == 1:
        deepclassifier = True
    else:
        deepclassifier = False

    normalized = temp

    params = {'nhid':nh,'nlayers':nl,'dropout':dr}
    
    pvalues = []
    d = np.load(folder_path + 'datafile.npy')
    for i in range(numfile):
        print '#iter: ' + str(i)
        datafile = folder_path  + 'datafile' + str(i) + '_' + str(dz) + '.csv'
        y = pd.read_csv(datafile,header = None)
        y = np.array(y).astype(np.float32)
        y = y[1::,1::]
        
        MCG = MIMIFY_GAN(y[:,0:dx],y[:,dx:dx+dy],y[:,dx+dy:dx+dy+dz],\
                 normalized=normalized,nthread = nthread, dim_N = dim_N, max_epoch = maxepoch, bsize = bsize,deep_classifier=deepclassifier,params = params)
        pvalues = pvalues + [MCG.CI_classify()] 
        print i,d[i],pvalues[-1]

        np.save(folder_path + rfile + '.npy' ,pvalues)