How does the sliding window method work?

[rosho]

Hi

I am not sure how the "Sliding window" method work.

Let's assume I have a dataset of number of logins by hour.

a) A window of 24hours to predict the next 24h?

b) A window of 24h to predict the next 1h?

Thank you

[grana_splunk · Accepted solution]

I would recommend you to check our new blog on MLTK 4.3 which has a new Smart assistant and an example on maximum number of logins. https://www.splunk.com/blog/2019/06/03/what-s-new-in-the-splunk-machine-learning-toolkit-4-3.html

[rosho]

The problem is that when I update Splunk MLTK, the metrics (ex, RMSE in the Forecasting showcase) no longer appear.

[grana_splunk]

Can you share your MLTK version? Also, did you checked Experiment History option. It saves your old execution results

[rosho]

4.2.0 shows the metrics

Then I upgraded to the latest one and now I do not see the metrics.

[grana_splunk]

Use the Experiment not showcase.

Click on Experiments--> Click on Forecasting Assistant and then your history will be saved.

Showcase is just to show and guide user and its for example purpose. Experiments will retain your previous history.

[grana_splunk]

Can you add more detail to your question? is it about forecasting or prediction?

[rosho]

forecasting. I want to forecast the number of logins.
But I have divided my dataset in TRAIN and TEST. So it will be prediction for the TEST.