import numpy as np
from convnet import ConvolutionalNetwork
from sklearn.cross_validation import train_test_split
from sklearn.metrics import classification_report
# Dataset
X = np.loadtxt('../feature/3grams_count_mc_features')
y = np.loadtxt('../data/tag_mc')
X -= X.min()
X /= X.max()
X = X.reshape(-1, 1, 8, 8)
y = np.asarray(y)
y = y.astype(np.int32)
X_train, X_test, y_train, y_test = train_test_split(X, y)
# Instanciation
convnet = ConvolutionalNetwork()
# Train
convnet.train(X_train, y_train)
# Report
preds = convnet.predict(X_test)
tags = y_test
print classification_report(tags, preds)
convnet = ConvolutionalNetwork()
#Training
convnet.train(X, y)
#Test dataset
X = np.loadtxt('test_data')
y = np.loadtxt('test_labels')
X, y = shuffle(X, y)
#Data normalization
X -= X.min()
X /= X.max()
#Data reshape
X = X.reshape(-1, 1, 8, 8)
y = np.asarray(y)
y = y.astype(np.int32)
#Predictions
predictions = convnet.predict(X)
#Report
print classification_report(y, predictions)
print 'Accuracy: ' + str(accuracy_score(y, predictions))