Python mainの例

コード例 #1

ファイル: pinterest_model.py プロジェクト: messiest/corgi-toolkit

def main():
    if os.path.exists(MODEL_FILE):
        logistic = joblib.load(MODEL_FILE)
    else:
        logistic = LogisticRegression()

    if not os.path.exists(DATA_FILE):
        feature_extraction.main('blog post', 'glamour')

    df = pd.read_csv(DATA_FILE)

    df = df.fillna(0)

    print(df.shape)

    # TODO (@messiest) y ~ 1 if y in top 25%, 0 else

    y = df['impact'].apply(lambda x: 0 if x < df['impact'].mean() else 1)

    x = df.iloc[:, 36:]

    print(x.shape, y.shape)

    x_train, x_test, y_train, y_test = train_test_split(x, y)
    logistic.fit(x_train, y_train)
    predictions = logistic.predict(x_test)
    print(classification_report(y_test, predictions))

    joblib.dump(logistic, MODEL_FILE)

コード例 #2

def model():
    if os.path.exists(MODEL_FILE):
        logistic = joblib.load(MODEL_FILE)
    else:
        logistic = LogisticRegression(penalty='l2')

    if not os.path.exists(DATA_FILE):
        feature_extraction.main('blog post', 'glamour')

    df = pd.read_csv(DATA_FILE)

    df = df.fillna(0)

    y = df['impact'].apply(lambda x: 0 if x < df['impact'].mean() else 1)
    x = df.iloc[:, 36:]

    x_train, x_test, y_train, y_test = train_test_split(x, y)
    logistic.fit(x_train, y_train)
    predictions = logistic.predict(x_test)
    print(classification_report(y_test, predictions))

    joblib.dump(logistic, MODEL_FILE)

    features = dict(zip(x.columns, list(logistic.coef_[0])))

    return features

コード例 #3

ファイル: predict_audio.py プロジェクト: felipemalbergier/api-tcc

def predict_audio(clf_and_scaler_folder, used_features, clf_to_use):
    output_path = config['OUTPUT_FOLDER']

    # From csv files get features from 1 second frames
    features = fex.main(output_path, used_features, clf_to_use)

    scaler_file = 'scaler.sav'
    scaler_file = clf_and_scaler_folder + '/' + scaler_file
    scaler = pickle.load(open(scaler_file, 'rb'))

    clf_file = 'finalized_model.sav'
    clf_file = clf_and_scaler_folder + '/' + clf_file
    clf = pickle.load(open(clf_file, 'rb'))

    scaled_data = scaler.transform(features)

    predicted = clf.predict(scaled_data)
    new_predicted = remove_outliers(predicted)

    new_predicted = list(new_predicted)

    likely_to_be_music = sum(new_predicted) / len(new_predicted)

    return (likely_to_be_music, new_predicted)

コード例 #4

ファイル: run.py プロジェクト: jbrowarczyk/jb-masters-thesis

import comp_dataset
import preproc
import feature_extraction
import experiment1
import experiment2
import experiment3
import experiment4

build_dataset.main()
preprocessing.main()
feature_extraction.main()
experiment1.main()
experiment2.main()
experiment3.main()
experiment4.main()

コード例 #5

ファイル: index.py プロジェクト: montyshyama/phishing-url-detection

from sklearn.externals import joblib
import feature_extraction

classifier = joblib.load('model/trained.pkl')

print('[+]Enter URL:')
url = input()
check = feature_extraction.main(url)
prediction = classifier.predict(check)
print(prediction)