Commit a7bd9d18 authored by mjennewine's avatar mjennewine
Browse files

partial fit new data

parent 01931896

ml.py

+104 −77
Original line number Diff line number Diff line
@@ -21,52 +21,35 @@ sgd = None 
nn = None

tree = None



def getPerformanceMetrics():

def preprocessing():

    global X, Y

    XYList = preprocess.loadXY()

    X = XYList[0]

    Y = XYList[1]





def testModels():

    if len(X) == 0:

        preprocessing()

    

    if tree == None:

        loadModels()

    

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



    pla_yPred = pla.predict(X_test)

    sgd_yPred = sgd.predict(X_test)

    nn_yPred = nn.predict(X_test)

    tree_yPred = tree.predict(X_test)

    

    accuracy = [accuracy_score(Y_test, pla_yPred),

                   accuracy_score(Y_test, sgd_yPred),

                   accuracy_score(Y_test, nn_yPred),

                   accuracy_score(Y_test, tree_yPred)]

    

    precision = [precision_score(Y_test, pla_yPred, average='macro'),

                   precision_score(Y_test, sgd_yPred, average='macro'),

                   precision_score(Y_test, nn_yPred, average='macro'),

                   precision_score(Y_test, tree_yPred, average='macro')]

        

    recall = [recall_score(Y_test, pla_yPred, average='macro'),

               recall_score(Y_test, sgd_yPred, average='macro'),

               recall_score(Y_test, nn_yPred, average='macro'),

               recall_score(Y_test, tree_yPred, average='macro')]

    

    f1 = [recall_score(Y_test, pla_yPred, average='macro'),

           recall_score(Y_test, sgd_yPred, average='macro'),

           recall_score(Y_test, nn_yPred, average='macro'),

           recall_score(Y_test, tree_yPred, average='macro')]

    pred_y = pla.predict(X)

    print('Perceptron (all data):')

    printMetrics(Y, pred_y)

    

    # accuracy = accuracy_score(y_actual, y_pred)

    # precision = precision_score(y_actual, y_pred, average='macro')

    # recall = recall_score(y_actual, y_pred, average='macro')

    # f1 = f1_score(y_actual, y_pred, average='macro')

    pred_y = sgd.predict(X)

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

    printMetrics(Y, pred_y)

    

    return accuracy, precision, recall, f1

    pred_y = nn.predict(X)

    print('\nNeural Network (all data):')

    printMetrics(Y, pred_y)

    

def printMetrics(y_actual, y_pred):

    print('Accuracy: ', accuracy_score(y_actual, y_pred))

    print('Precision:', precision_score(y_actual, y_pred, average='macro'))

    print('Recall:   ', recall_score(y_actual, y_pred, average='macro'))

    print('F1:       ', f1_score(y_actual, y_pred, average='macro'))

    pred_y = tree.predict(X)

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

    printMetrics(Y, pred_y)





def binarize(value):
@@ -84,11 +67,11 @@ def binarize_list(data): 
    return result





def preprocessing():

    global X, Y

    XYList = preprocess.loadXY()

    X = XYList[0]

    Y = XYList[1]

def printMetrics(y_actual, y_pred):

    print('Accuracy: ', accuracy_score(y_actual, y_pred))

    print('Precision:', precision_score(y_actual, y_pred, average='macro'))

    print('Recall:   ', recall_score(y_actual, y_pred, average='macro'))

    print('F1:       ', f1_score(y_actual, y_pred, average='macro'))





def trainModels():
@@ -120,7 +103,8 @@ def trainModels(): 
    print('\nPerceptron (testing data):')

    printMetrics(Y_test, pred_y_test)

    

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

    pla = pla.best_estimator_

    dump(pla, 'pla.joblib')

    print('==================================================\n\n')
@@ -146,7 +130,8 @@ def trainModels(): 
    print('\nSGD (testing data):')

    printMetrics(Y_test, pred_y_test)

    

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

    sgd = sgd.best_estimator_

    dump(sgd, 'sgd.joblib')

    print('==================================================\n\n')
@@ -172,7 +157,8 @@ def trainModels(): 
    print('\nNeural Network (testing data):')

    printMetrics(Y_test, pred_y_test)

    

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

    nn = nn.best_estimator_

    dump(nn, 'nn.joblib')

    print('==================================================\n\n')
@@ -208,30 +194,6 @@ def loadModels(): 
        tree = load('tree.joblib')





def testModels():

    if len(X) == 0:

        preprocessing()

    

    if tree == None:

        loadModels()

    

    pred_y = pla.predict(X)

    print('Perceptron (all data):')

    printMetrics(Y, pred_y)

    

    pred_y = sgd.predict(X)

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

    printMetrics(Y, pred_y)

    

    pred_y = nn.predict(X)

    print('\nNeural Network (all data):')

    printMetrics(Y, pred_y)

    

    pred_y = tree.predict(X)

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

    printMetrics(Y, pred_y)



#### not finished 

def makePrediction(message):

    if tree == None:

        loadModels()
@@ -257,19 +219,84 @@ def makePrediction(message): 
    return str(pred_y.count(1)) + ' out of 4 models predict this message to be spam.'





# declaring, training, fitting each algorithm

def mlinit():

    pass

def getPerformanceMetrics():

    if len(X) == 0:

        preprocessing()

    

    if tree == None:

        loadModels()

        

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



    pla_yPred = pla.predict(X_test)

    sgd_yPred = sgd.predict(X_test)

    nn_yPred = nn.predict(X_test)

    tree_yPred = tree.predict(X_test)

    

    accuracy = [accuracy_score(Y_test, pla_yPred),

                   accuracy_score(Y_test, sgd_yPred),

                   accuracy_score(Y_test, nn_yPred),

                   accuracy_score(Y_test, tree_yPred)]

    

    precision = [precision_score(Y_test, pla_yPred, average='macro'),

                   precision_score(Y_test, sgd_yPred, average='macro'),

                   precision_score(Y_test, nn_yPred, average='macro'),

                   precision_score(Y_test, tree_yPred, average='macro')]

        

    recall = [recall_score(Y_test, pla_yPred, average='macro'),

               recall_score(Y_test, sgd_yPred, average='macro'),

               recall_score(Y_test, nn_yPred, average='macro'),

               recall_score(Y_test, tree_yPred, average='macro')]

    

    f1 = [f1_score(Y_test, pla_yPred, average='macro'),

           f1_score(Y_test, sgd_yPred, average='macro'),

           f1_score(Y_test, nn_yPred, average='macro'),

           f1_score(Y_test, tree_yPred, average='macro')]

    

    # accuracy = accuracy_score(y_actual, y_pred)

    # precision = precision_score(y_actual, y_pred, average='macro')

    # recall = recall_score(y_actual, y_pred, average='macro')

    # f1 = f1_score(y_actual, y_pred, average='macro')

    

    return accuracy, precision, recall, f1



# partial fit the new prediction data to each algorithm

def partialFitNewData(message, label):

    pass

    before = getPerformanceMetrics()

    global pla, sgd, nn, tree

    

    x_new = [preprocess.extract(message)]

    y_new = [0 if label == 'ham' else 1]

    

    print('Updating Perceptron Model...')

    pla.partial_fit(x_new, y_new)

    #dump(pla, 'pla.joblib')

    

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

    sgd.partial_fit(x_new, y_new)

    #dump(sgd, 'sgd.joblib')

    

    print('Updating Neural Network...')

    nn.partial_fit(x_new, y_new)

    #dump(nn, 'nn.joblib')

    

    print('Retraining Decision Tree...')

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

    tree = DecisionTreeClassifier()

    tree.fit(np.concatenate((X_train, x_new), axis=0), np.concatenate((Y_train, y_new), axis=0))

    #dump(tree, 'tree.joblib')

    

    after = getPerformanceMetrics()

    print(np.subtract(after, before))

    return 'Models updated successfully'



def switchCurrentAlgorithm():

    pass



# def getPerformanceMetrics():

#     pass



def getSampleMessages():

    pass



# declaring, training, fitting each algorithm

def mlinit():

    preprocessing()

    loadModels()
 
 No newline at end of file

server.py

+0 −3
Original line number Diff line number Diff line
@@ -59,7 +59,6 @@ def getTopAlgorithms(): 
    name = ["Perceptron", "Stochastic Gradient Descent", "Neural Network", "Decision Tree"]

    # accuracy, precision, recall, f1 = ml.getPerformanceMetrics()

    result = ml.getPerformanceMetrics()

    print(result)

    accuracy = result[0]

    precision = result[1]

    recall = result[2]
@@ -84,7 +83,6 @@ def getTopAlgorithms(): 
        "accuracy": accuracy

    }





    return val



# TODO make this get the actual top feature words
@@ -101,5 +99,4 @@ def getTopFiveWords(): 
# Run the app!

if __name__ == '__main__':

    app.run(port = 8080)

    ml.preprocessing()

    ml.mlinit()
 
 No newline at end of file