plt.ylabel('Cost')
plt.show()
# fig.savefig('linReg_regularization_comparison.eps', format='eps')
# predict and eval on test set
print '... predicting ...'
# add column of ones to data to account for the bias:
X_test = add_intercept(X_test)
print X_test.shape
pred = list()
for cx in X_test:
pred.append(predict(cx))
y_hat = np.array(pred, dtype=float)
RMSE, pcorr, error_per_song, mean_per_song = evaluate1d(
y_test, y_hat, tst_song)
All_stack = np.hstack((error_per_song, mean_per_song))
print ' Error per song (ar/val) Mean_per_song (ar/val) :\n'
print(All_stack)
print '\n'
print 'song_id :'
print(song_id_tst)
print '\n'
#print('Error per song: \n', Error_per_song)
print('sklearn --> arrousal : %.4f, valence : %.4f\n'
'Pearson Corr --> arrousal : %.4f, valence : %.4f \n' %
(RMSE[0], -1., pcorr[0][0], -1)
# % (RMSE[0],RMSE[1],pcorr[0][0], pcorr[1][0])
print res.summary()
# cf http://statsmodels.sourceforge.net/devel/mixed_linear.html
# md = smf.mixedlm(y_train, X_train, groups=data["Pig"])
# mdf = md.fit()
# print(mdf.summary())
pred = res.predict(X_test)
# print pred
all_fold_pred.append(pred)
all_fold_y_test.append(y_test.tolist())
y_hat = np.array(pred, dtype=float)
RMSE, pcorr, error_per_song, mean_per_song = evaluate1d(y_test, y_hat, id_test.shape[0])
# All_stack =np.hstack(( error_per_song, mean_per_song ))
# print' Error per song (ar/val) Mean_per_song (ar/val) :\n'
# print(All_stack)
# print '\n'
#
# print'song_id :'
# print(id_test)
# print '\n'
# #print('Error per song: \n', Error_per_song)
print(
'sklearn --> arrousal : %.4f, valence : %.4f\n'
'Pearson Corr --> arrousal : %.4f, valence : %.4f \n'
% (RMSE[0], -1. , pcorr[0][0], -1)
plt.ylabel("Cost")
plt.show()
# fig.savefig('linReg_regularization_comparison.eps', format='eps')
# predict and eval on test set
print "... predicting ..."
# add column of ones to data to account for the bias:
X_test = add_intercept(X_test)
print X_test.shape
pred = list()
for cx in X_test:
pred.append(predict(cx))
y_hat = np.array(pred, dtype=float)
RMSE, pcorr, error_per_song, mean_per_song = evaluate1d(y_test, y_hat, tst_song)
All_stack = np.hstack((error_per_song, mean_per_song))
print " Error per song (ar/val) Mean_per_song (ar/val) :\n"
print (All_stack)
print "\n"
print "song_id :"
print (song_id_tst)
print "\n"
# print('Error per song: \n', Error_per_song)
print (
"sklearn --> arrousal : %.4f, valence : %.4f\n"
"Pearson Corr --> arrousal : %.4f, valence : %.4f \n" % (RMSE[0], -1.0, pcorr[0][0], -1)
# % (RMSE[0],RMSE[1],pcorr[0][0], pcorr[1][0])
plt.show()
# fig.savefig('linReg_regularization_comparison.eps', format='eps')
# predict and eval on test set
print '... predicting ...'
pred = list()
for cx in X_test:
pred.append(predict(cx))
all_fold_pred.append(pred)
all_fold_y_test.append(y_test.tolist())
y_hat = np.array(pred, dtype=float)
RMSE, pcorr, error_per_song, mean_per_song = evaluate1d(
y_test, y_hat, id_test.shape[0])
# All_stack =np.hstack(( error_per_song, mean_per_song ))
# print' Error per song (ar/val) Mean_per_song (ar/val) :\n'
# print(All_stack)
# print '\n'
#
# print'song_id :'
# print(id_test)
# print '\n'
# #print('Error per song: \n', Error_per_song)
print('sklearn --> arrousal : %.4f, valence : %.4f\n'
'Pearson Corr --> arrousal : %.4f, valence : %.4f \n' %
(RMSE[0], -1., pcorr[0][0], -1)
# % (RMSE[0],RMSE[1],pcorr[0][0], pcorr[1][0])
