Use logistic regression on time series data

Hi everyone,
Recently, I’m seeking a model solution to make binary decision (buy or sell) based on time series data. I’m considering ARIMA, Auto-Regressive or RNN. However, they only forecast and make numeric predictions, not decisions like buy or sell. Does anyone now how to integrate logistic regression into these kinds of models? Or should I change to using only basic neural network for logistic regression?

1 Like

What’s the problem in making sigmoid as output layer activation function and adjusting the loss function accordingly for RNN based approach?

1 Like

RNN is one approach that I’m considering, I haven’t deployed it. I’m asking about making a sigmoid layer for ARIMA and Auto-Regressive.

1 Like

I haven’t used ARIMA for classification. That said, how about using the predicted value of the target before sigmoid as a feature to train a classifier? Here’s one that might help.