Commit 8c431c32 authored by mjennewine's avatar mjennewine
Browse files

better prediction + finally pushed X.joblib

parent f9dc158e

X0.joblib

0 → 100644
+76.3 MiB

File added.

No diff preview for this file type.

View file

X1.joblib

0 → 100644
+81.5 MiB

File added.

No diff preview for this file type.

View file

ml.py

+62 −25
Original line number Diff line number Diff line
@@ -5,6 +5,7 @@ from sklearn.linear_model import Perceptron, SGDClassifier 
from sklearn.neural_network import MLPClassifier

from sklearn.tree import DecisionTreeClassifier

from joblib import dump, load

import numpy as np

import os



from warnings import simplefilter
@@ -16,7 +17,7 @@ import preprocess 
X = []

Y = []

pla = None

sgc = None

sgd = None

nn = None

tree = None
@@ -28,6 +29,21 @@ def printMetrics(y_actual, y_pred): 
    print('F1:       ', f1_score(y_actual, y_pred, average='macro'))





def binarize(value):

    if value > 0.5:

        return 1

    else:

        return 0





def binarize_list(data):

    result = np.zeros(len(data)).astype(np.int)

    for i in range(len(data)):

        if data[i] > 0.5:

            result[i] = 1

    return result





def preprocessing():

    global X, Y

    XYList = preprocess.loadXY()
@@ -40,9 +56,9 @@ def trainModels(): 
    

    X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2, random_state=0)

    

    global pla, sgc, nn, tree

    global pla, sgd, nn, tree

    

    print('\n==================================================\n')

    print('Training Perceptron Model...')

    

    ### train perceptron

    # 'n_jobs': [16]
@@ -52,6 +68,7 @@ def trainModels(): 
    pred_y_train = pla.predict(X_train)

    pred_y_test = pla.predict(X_test)

    

    print('==================================================')

    print('Perceptron CV parameters:')

    print('Best: ', pla.best_params_)

    accuracy = pla.cv_results_['mean_test_score']
@@ -64,21 +81,24 @@ def trainModels(): 
    printMetrics(Y_test, pred_y_test)

    

    dump(pla.best_estimator_, 'pla.joblib')

    print('\n==================================================\n')

    print('==================================================\n\n')

    

    

    ### train logistic regression

    # 'n_jobs': [16]

    print('Training Stochastic Gradient Descent Model...')

    

    params = {'alpha': [.00001, .0001, .001, .01, .1]}

    sgc = GridSearchCV(SGDClassifier(), params, cv=5)

    sgc.fit(X_train, Y_train)

    pred_y_train = sgc.predict(X_train)

    pred_y_test = sgc.predict(X_test)

    sgd = GridSearchCV(SGDClassifier(), params, cv=5)

    sgd.fit(X_train, Y_train)

    pred_y_train = sgd.predict(X_train)

    pred_y_test = sgd.predict(X_test)

    

    print('==================================================')

    print('Stochastic Gradient Descent CV parameters:')

    print('Best: ', sgc.best_params_)

    accuracy = sgc.cv_results_['mean_test_score']

    for acc, param in zip(accuracy, sgc.cv_results_['params']):

    print('Best: ', sgd.best_params_)

    accuracy = sgd.cv_results_['mean_test_score']

    for acc, param in zip(accuracy, sgd.cv_results_['params']):

        print("%0.5f - %r"% (acc, param))

    

    print('\nSGD (training data):')
@@ -86,18 +106,21 @@ def trainModels(): 
    print('\nSGD (testing data):')

    printMetrics(Y_test, pred_y_test)

    

    dump(sgc.best_estimator_, 'sgc.joblib')

    print('\n==================================================\n')

    dump(sgd.best_estimator_, 'sgd.joblib')

    print('==================================================\n\n')

    

    

    ### train neural network

    #params = {'hidden_layer_sizes': [(500,), (1000,), (1500,)], 'alpha': [.0001, .001, .01]}

    print('Training Neural Network...')

    

    params = {'hidden_layer_sizes': [(100,)], 'alpha': [.001]}

    nn = GridSearchCV(MLPClassifier(), params, cv=5)

    nn.fit(X_train, Y_train)

    pred_y_train = nn.predict(X_train)

    pred_y_test = nn.predict(X_test)

    

    print('==================================================')

    print('Neural Network CV parameters:')

    print('Best: ', nn.best_params_)

    accuracy = nn.cv_results_['mean_test_score']
@@ -110,34 +133,37 @@ def trainModels(): 
    printMetrics(Y_test, pred_y_test)

    

    dump(nn.best_estimator_, 'nn.joblib')

    print('\n==================================================\n')

    print('==================================================\n\n')

    

    

    ### train decision tree

    print('Training Decision Tree...')

    

    tree = DecisionTreeClassifier()

    tree.fit(X_train, Y_train)

    pred_y_train = tree.predict(X_train)

    pred_y_test = tree.predict(X_test)

    

    print('==================================================')

    print('Decision Tree (training data):')

    printMetrics(Y_train, pred_y_train)

    print('\nDecision Tree (testing data):')

    printMetrics(Y_test, pred_y_test)

    

    dump(tree, 'tree.joblib')

    print('\n==================================================\n')

    print('==================================================\n\n')





def loadModels():

    global pla, sgc, nn, tree

    global pla, sgd, nn, tree

    

    if not os.path.isfile('pla.joblib') or not os.path.isfile('sgc.joblib') or not os.path.isfile('nn.joblib') or not os.path.isfile('tree.joblib'):

    if not os.path.isfile('pla.joblib') or not os.path.isfile('sgd.joblib') or not os.path.isfile('nn.joblib') or not os.path.isfile('tree.joblib'):

        print('Training models...')

        trainModels()

    else:

        print('Loading models from file...')

        pla = load('pla.joblib')

        sgc = load('sgc.joblib')

        sgd = load('sgd.joblib')

        nn = load('nn.joblib')

        tree = load('tree.joblib')
@@ -153,7 +179,7 @@ def testModels(): 
    print('Perceptron (all data):')

    printMetrics(Y, pred_y)

    

    pred_y = sgc.predict(X)

    pred_y = sgd.predict(X)

    print('\nStochastic Gradient Descent (all data):')

    printMetrics(Y, pred_y)
@@ -170,13 +196,25 @@ def makePrediction(message): 
    if tree == None:

        loadModels()

    

    pred_y = tree.predict([preprocess.extract(message)])

    #predicted label

    pred_y = []

    pred_y.append(pla.predict([preprocess.extract(message)])[0])

    pred_y.append(sgd.predict([preprocess.extract(message)])[0])

    pred_y.append(nn.predict([preprocess.extract(message)])[0])

    pred_y.append(tree.predict([preprocess.extract(message)])[0])

    print(pred_y, '\n')

    

    """

    #confidence values

    pred_y = []

    pred_y.append(pla.decision_function([preprocess.extract(message)])[0])

    pred_y.append(sgd.decision_function([preprocess.extract(message)])[0])

    pred_y.append(nn.predict_proba([preprocess.extract(message)])[0][1])

    pred_y.append(tree.predict_proba([preprocess.extract(message)])[0][1])

    print(pred_y)

    """

    

    if pred_y[0] == 1:

        return 'spam'

    else:

        return 'ham'

    return str(pred_y.count(1)) + ' out of 4 modes predict this message to be spam.'





# declaring, training, fitting each algorithm
@@ -188,4 +226,3 @@ def partialFitNewData(): 


def switchCurrentAlgorithm():

    pass

nn.joblib

−310 B (3.07 MiB)

File changed.

No diff preview for this file type.

View original file
View changed file

pla.joblib

(8.91 KiB)

File changed.

No diff preview for this file type.

View original file
View changed file