mlatoz

Recurrent Neural Networks (RNNs)

Evaluating the RNN

Hi guys,

as seen in the practical lectures, the RNN we built was a regressor. Indeed, we were dealing with Regression because we were trying to predict a continuous outcome (the Google Stock Price). For Regression, the way to evaluate the model performance is with a metric called RMSE (Root Mean Squared Error). It is calculated as the root of the mean of the squared differences between the predictions and the real values.

However for our specific Stock Price Prediction problem, evaluating the model with the RMSE does not make much sense, since we are more interested in the directions taken by our predictions, rather than the closeness of their values to the real stock price. We want to check if our predictions follow the same directions as the real stock price and we don’t really care whether our predictions are close the real stock price. The predictions could indeed be close but often taking the opposite direction from the real stock price.

Nevertheless if you are interested in the code that computes the RMSE for our Stock Price Prediction problem, please find it just below:

import math
from sklearn.metrics import mean_squared_error
rmse = math.sqrt(mean_squared_error(real_stock_price, predicted_stock_price))

Then consider dividing this RMSE by the range of the Google Stock Price values of January 2017 (that is around 800) to get a relative error, as opposed to an absolute error. It is more relevant since for example if you get an RMSE of 50, then this error would be very big if the stock price values ranged around 100, but it would be very small if the stock price values ranged around 10000.

Enjoy Deep Learning!

Improving the RNN

Hi guys,

here are different ways to improve the RNN model:

Getting more training data: we trained our model on the past 5 years of the Google Stock Price but it would be even better to train it on the past 10 years.
Increasing the number of timesteps: the model remembered the stock prices from the 60 previous financial days to predict the stock price of the next day. That’s because we chose a number of 60 timesteps (3 months). You could try to increase the number of timesteps, by choosing for example 120 timesteps (6 months).
Adding some other indicators: if you have the financial instinct that the stock price of some other companies might be correlated to the one of Google, you could add this other stock price as a new indicator in the training data.
Adding more LSTM layers: we built a RNN with four LSTM layers but you could try with even more.
Adding more neurons in the LSTM layers: we highlighted the fact that we needed a high number of neurons in the LSTM layers to respond better to the complexity of the problem and we chose to include 50 neurons in each of our 4 LSTM layers. You could try an architecture with even more neurons in each of the 4 (or more) LSTM layers.

Enjoy Deep Learning!

«Previous