[Python] How to predict nonlinear, uncorrelated features with few data points using machine...

So I am working on a prediction task where my goal is to predict the volume of the reservoir and current level. I have some SITREPs data but it isn't consistent and isn't enough to train a reliable model (i.e. only 26 rows).

My best bet is to rely on remote sensing so I am calculating surface area using satellite imagery which gave me a timeseries for 10 years on a weekly basis. I then interpolated the data to get daily values and got 2k rows which I think is good enough.

I tried FBProphet model for timeseries forecasting and it did forecasted but the values doesn't seem to be reliable.

I am trying to use regression models to predict the surface area with other variables such as rainfall but the correlation between the two values is negligible. How can I find a workaround for this problem?

I have tried using AI but it's not working that's why I am posting it here.

Continue reading...