Machine learning in economy is very important, and one of its application is predicting the behavior of the stock market.
How can you do that, the idea is creating a model that reads the previous graph of the market and gives the predicted behavior as output.
To do so, one cannot use the simple Tensorflow Dense layers. The more sophisticated Long Short Term Memory layers, capable of reading entire sequences of data and used in important fields like speech and handwriting recognition, are required.
You can find a basic model that, from the last 60 days, is able to accurately predict the stock value of the following day on my github page (I will provide a link below). The ideal would be to predict more than one day, but there’s time for the project to be improved.
https://github.com/mattiagiuri/machine-learning-projects/tree/main/Stock%20Prediction