raw_test_prob = model.predict(test_df)
test_pred = np.expm1(raw_test_prob)
return results.history, test_pred, val_idx, raw_val_prob, raw_test_prob
if __name__ == '__main__':
start_time = time.time()
preprocess_dict = preprocess_csv()
(train_ids, test_ids, processed_train_df, processed_test_df) = [
preprocess_dict[key] for key in
['train_ids', 'test_ids', 'processed_train', 'processed_test']
]
folds = split_to_folds(processed_train_df,
num_folds=NUM_FOLDS,
seed=SEED,
shuffle=False)
all_fold_results = []
all_fold_preds = []
with Pool(NUM_FOLDS) as p:
combined_results = p.starmap(
train_fold, ((curr_fold, processed_train_df, processed_test_df)
for curr_fold in folds))
losses = np.mean([x[0]['loss'] for x in combined_results], axis=0)
val_losses = np.mean([x[0]['val_loss'] for x in combined_results], axis=0)
print('Loss: {}'.format(losses))
print('Val loss: {}'.format(val_losses))
plt.plot(losses, label='train_losses')
plt.plot(val_losses, label='val_losses')
validation_data=(val_x, val_y),
epochs=NUM_EPOCHS)
test_pred = model.predict(test_df)
return model, results.history, test_pred
if __name__ == '__main__':
preprocess_dict = preprocess_csv()
(train_ids, test_ids, processed_train_df, processed_test_df) = [
preprocess_dict[key] for key in
['train_ids', 'test_ids', 'processed_train', 'processed_test']
]
folds = split_to_folds(processed_train_df, NUM_FOLDS, SEED)
all_fold_results = []
all_fold_preds = []
for curr_fold in folds:
_, curr_results, curr_pred = train_fold(
build_dense_model(num_features=processed_train_df.shape[1] - 1),
curr_fold, processed_train_df, processed_test_df)
all_fold_results.append(curr_results)
all_fold_preds.append(curr_pred)
print('Val loss: {}'.format(
np.mean([x['val_loss'] for x in all_fold_results], axis=0)))
print('Val RMSLE: {}'.format(
np.mean([x['val_rmsle'] for x in all_fold_results], axis=0)))
