Forecast IT

Forecasting and modelling time-series based on the past observations of climate data.

There are seventh time-series data features:

- Gdd, is Growing degree days. Used for example estimate when fruits are ripe. 

- pmm, is precipitation in milimeters

- rad, is radiation energy from the sun in kW/m^2

- rad_h, is sun hours for a specific point. How many hours of sun are received for a specific day and point?

- tmax and tmin: Temperature max and min for a day and location

- tmean: mean temperature . 

 

The main research question is: can we predict gdd based on the past observations of the 3 previous day? What error do we get?

Apparently, 11% of error with Random Forest Regressor.

Vdeo: https://vimeo.com/user104317687