SGDClassifier.py

import csv
import numpy as np
from sklearn import linear_model
from sklearn.metrics.classification import accuracy_score
from sklearn.metrics import classification_report


def loaddata(filename,instanceCol):
    file_reader = csv.reader(open(filename,'r'),delimiter=',')
    x = []
    y = []
    for row in file_reader:
        x.append(row[0:instanceCol])
        y.append(row[-1])
    return np.array(x[1:]).astype((np.float32)), np.array(y[1:]).astype(np.int)


def fractal_modeldata(filename, inc):
    print(filename)
    X, Y = loaddata(filename, inc)
    np.random.seed(13)
    indices = np.random.permutation(2038)
    test_size = int(0.3 * len(indices))
    X_train = X[indices[:-test_size]]
    Y_train = Y[indices[:-test_size]]
    X_test = X[indices[-test_size:]]
    Y_test = Y[indices[-test_size:]]
    classifier = linear_model.SGDClassifier(max_iter=10, loss='log')
    classifier.fit(X_train, Y_train)
    Y_pred = classifier.predict_proba(X_test)
    print(accuracy_score(Y_test, Y_pred)*100)
    print(classification_report(Y_test, Y_pred))


if __name__ == '__main__':
    fractal_modeldata('D:\\Databases\\Steganalysis\\Dataset\\LogFBank\\LogFBank-Features-hide4pgp-100.csv', 27)
    fractal_modeldata('D:\\Databases\\Steganalysis\\Dataset\\FBank\\FBank-Features-hide4pgp-100.csv', 27)
    fractal_modeldata('D:\\Databases\\Steganalysis\\Dataset\\Fractal\\Fractal-Features-hide4pgp-100.csv', 27)
    fractal_modeldata('D:\\Databases\\Steganalysis\\Dataset\\MFCC\\MFCC-Features-hide4pgp-100.csv', 27)
    fractal_modeldata('D:\\Databases\\Steganalysis\\Dataset\\LPC\\lpc-Features-hide4pgp-100.csv', 27)
    fractal_modeldata('D:\\Databases\\Steganalysis\\Dataset\\DeltaMFCC\\deltaMFCC-Features-hide4pgp-100.csv', 15)