program.py

import model
import argparse


class Train(argparse.Action):
    def __call__(self, parser, namespace, values, option_string):
        result = dict()
        # try:
        #     val = float(values)
        # except Warning:
        #     print('Wrong argument exception, value error')
        #     return
        if values is None:
            print('Training ....')
            result = model.train()
        elif 0.1 <= float(values) < 1:
            print('Training ....')
            result = model.train(float(values))
        else:
            print('Wrong train size, train size should be > 0.5 and < 1')
            return
        print('R - Square:')
        for item in result:
            print(item, result[item])
        setattr(namespace, self.dest, values)

# Create command
parser = argparse.ArgumentParser(description='Data mining program project to predict student\'s grades')

parser.add_argument('-t', '--train', action=Train, metavar='', nargs="?", help='Train model, can specific train size,'
                                                                               'default size is 0.7')

parser.add_argument('-p', '--predict', action='store', metavar='',
                    help='-p <sample> Predict student\'s grade')

parser.add_argument('-alg', '--algorithm', dest='alg', default='lr', choices=['lr', 'rf'],
                    metavar='', help='Choose the algorithm to predict student\'s grade \'lr\' is linear regression, '
                                     '\'rf\' is random forest ex: python program.py -p 12,13,0,0,0,4 -alg rf')

parser.add_argument('-m', '--metrics', dest='metrics', action='store_true',
                    help='Evaluate mae(Mean absolute error) and rmse(Root mean squared error) of 2 models')
parser.add_argument('-r', '--rsquare', dest='r', action='store_true', help='Print the R - Square (R^2) of 2 models')
parser.add_argument('-f', '--formula', dest='formula', action='store_true', help='Print linear regression formula')
args = parser.parse_args()


def cal_accuracy():
    pass


def compare_accuracy():
    pass


def predict_value(arr):
    sample = list()
    try:
        arr = [float(x) for x in arr.split(',')]
        sample.append(arr)
    except ValueError:
        print('Wrong sample, please read usage for more details')
        return
    if args.alg:
        if args.alg == 'lr':
            print('Linear regression:')
            predict = model.predict_grade(sample, 'random')
            print('Predict grade: {}'.format(predict[0].round(2)))
        elif args.alg == 'rf':
            predict = model.predict_grade(sample, 'linear')
            print('Random forest:')
            print('Predict grade: {}'.format(predict[0].round(2)))

        else:
            print('Wrong alg {}, please read usage for more information'.format(args.alg))


if args.predict:
    predict_value(args.predict)
elif args.metrics:
    metrics = model.evaluate_metrics()
    lr = metrics['lr']
    rf = metrics['rf']
    print('Model metrics:')
    print('Linear regression:   ')
    print('mae: {}   rmse: {}'.format(lr['mae'], lr['rmse']))
    print('Random forest:')
    print('mae: {}   rmse: {}'.format(rf['mae'], rf['rmse']))
elif args.r:
    accuracy = model.r_square()
    print('R - Square value for 2 models:')
    for item in accuracy.keys():
        print('{} : {}'.format(item, accuracy[item]))
elif args.formula:
    formula = model.formula()
    print("Formula of the linear regression model")
    print(formula)