Deans-bike-lending-prediction/src/3-train.py

# -*- coding: utf-8 -*-
"""
@author: Jessica Cabral
"""
############################################################################
#       Train Script
############################################################################

import sys
import os
import numpy as np
import pandas as pd
from datetime import datetime

from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestRegressor
from sklearn.linear_model import LinearRegression
from sklearn.linear_model import Ridge
from sklearn.linear_model import Lasso
from sklearn.linear_model import ElasticNet
from sklearn.tree import DecisionTreeRegressor
from sklearn.svm import SVR # Support Vector Regression

from sklearn.model_selection import cross_validate

try:
    import cPickle as pickle
except ImportError:
    import pickle

##############################
#    Get Sys Paths
##############################
#if len(sys.argv) != 3:
#    sys.stderr.write('Arguments error. Usage:\n')
#    sys.stderr.write('\tpython train.py data/processed models\n')
#    sys.exit(1)

PATH_PROCESSED_DATA = sys.argv[1] #'../data/processed'
PATH_MODEL = sys.argv[2] #'../models'

if not os.path.exists(PATH_MODEL):
    os.makedirs(PATH_MODEL)
    
# Test data set split ratio
split = 0.33
seed = 201909

###############################
##    Create Folder model version
###############################
#try:
#    model_version = len(next(os.walk(PATH_MODEL))[1])
#    if model_version is None:
#        model_version = 0
#except StopIteration :
#    model_version = 0
#    
#model_version_name = 'modelv{}-{}'.format(model_version+1, datetime.now().strftime("%d%m%Y-%H%M%S"))
#model_version_path = r'{}/{}'.format(PATH_MODEL,model_version_name) 
#
#if not os.path.exists(model_version_path):
#    os.makedirs(model_version_path)
#
#print('Versao do modelo: {}'.format(model_version_name))
##############################
#    Features and Target
##############################
bike_processed = pd.read_csv(PATH_PROCESSED_DATA+'/bikes_processed.tsv', sep='\t')

X = bike_processed[bike_processed.columns.difference(['cnt', 'dteday'])].values
y = bike_processed['cnt'].values

##############################
#    Train Test Split
##############################
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=split, random_state=seed)

X_train_final = pd.DataFrame(X_train)
X_train_final['target'] = y_train

X_test_final = pd.DataFrame(X_test)
X_test_final['target'] = y_test 

#with open(os.path.join(model_version_path, 'train.tsv'), 'w', encoding='utf8') as fd_out_train:
with open(os.path.join(PATH_MODEL, 'train.tsv'), 'w', encoding='utf8') as fd_out_train:
    fd_out_train.write(X_train_final.to_csv(sep='\t', index=False))

#with open(os.path.join(model_version_path, 'test.tsv'), 'w', encoding='utf8') as fd_out_train:
with open(os.path.join(PATH_MODEL, 'test.tsv'), 'w', encoding='utf8') as fd_out_train:
    fd_out_train.write(X_test_final.to_csv(sep='\t', index=False))    
    
##############################
#    Train Model
##############################
print('Input matrix size {}'.format(bike_processed.shape))
print('X_train_final matrix size {}'.format(X_train_final.shape))
print('X_test_final matrix size {}'.format(X_test_final.shape))

X = X_train_final[X_train_final.columns.difference(['target'])].values
y= X_train_final['target'].values

regr = RandomForestRegressor(max_depth=2, n_estimators=100, random_state=seed)
regr.fit(X, y)  

print('Modelo treinado com sucesso!')

##############################
#    Save Model
##############################
    
#with open(os.path.join(model_version_path, '{}.pkl'.format(model_version_name)), 'wb') as fd:
with open(os.path.join(PATH_MODEL, 'model.pkl'), 'wb') as fd:
    pickle.dump(regr, fd)
    
#print('Modelo salvo em: {}'.format(model_version_path, '{}.pkl'.format(model_version_name)))
    print('Modelo salvo em: {}'.format(PATH_MODEL, 'model.pkl'))