Apple Stock Price Prediction using LSTM | Multivariate Time Series Forecasting using Deep Learning

[@yashsinghal7471]

Again very helpful.. Thank you!

[@KevinMiller-v7h]

Very interesting and informative video. One question? What would you change if you wanted to predict, say 3 days out? Just edit the 'create_sequence' function to set y values to 3 days out?

[@InsightsByRish]

Thanks. Yes, to predict 3 days out, you just need to modify the create_sequence function so that the y values point to the data 3 days ahead instead of the next day. The rest of the process stays the same.

[@rafabobo9593]

Thank you very much

[@hardik4417]

thank you for providing such content

[@InsightsByRish]

Glad you liked it!

[@Anthony-o1b2j]

Excellent video, thank you for sharing! I have a question, you're predicting Open, Close, Min, Max, Adj Close and Volume at the same time and get values for each one of these dimensions, that's what is called multivariate time series forecasting. But, when the LSTM is running, does it consider Close, Min, Max, Adj Close and Volume of the given period to predict the Open value? I mean are we predicting Open, Close, Min, Max, Adj Close and Volume INDEPENDENTLY or does the algorithm considers all the other dimensions to predict one of them?

[@InsightsByRish]

Hi @Anthony-o1b2j, thank you for your kind words! I'm glad you found the video helpful. To answer your question: Yes, when using an LSTM for multivariate time series forecasting, the model considers all features together. It doesn’t predict each feature independently. Instead, it takes into account the relationships and interdependencies between all the features to make predictions.

[@prawncracker1304]

Shouldn't the prediction line be out ahead of the actual line? I would expect to see the red line be further along in the x axis to show the prediction for a price that doesn't yet exist.

[@InsightsByRish]

The thing that you're saying will only happen if we are making predictions on new data or for further intervals of time. But here we are making predictions for existing data that was already present in X_test. So the red line (predicted values) will remain on top of blue line (actual values) as the predicted values are getting aligned with the existing targets present in y_test.