I have done feature engineering on a single variable time series data (spare parts usage), then I turn the time series data into supervised machine learning problem. I have trained and test on the transformed new dataset. My question is how to apply the model to predict one time step, multi time steps into the future? All features are calculated over prior data points, such as lag features, moving window statistics, such as max, min, mean, median over for example, last 5, 10, 15 days.

If you have working examples in Python or any white papers with example, that will be great. Look forward to your help Shaun

  • $\begingroup$ Thanks, Ben. Where can I upload an image with my sample data? Thanks $\endgroup$ – heroadu2019 Dec 7 '19 at 10:40

As your question is quite vague I just throw in some tools: MCMC, Kalman, LSTM, ARIMA, one-step-ahead and many more. Have a look at this to get started.

Can you please provide a part of your data? There are a lot of tools but it depends a lot on your situation and data.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.