search

e-sensing / sits

Satellite image time series in R
https://e-sensing.github.io/sitsbook/
GNU General Public License v2.0
477 stars 77 forks source link

sits_classify with XGBoost doesn't crop the ROI #1227

Open EstefaPizarro opened 1 week ago

EstefaPizarro commented 1 week ago

Describe the requested improvement When applying the X function, using XGBoost as a training model, the tiles are not cropped to the study area provided by the ROI parameter. The tiles are cropped to the study area when this process is performed with a Random Forest model. For the particular case of the Chilean machine, a test was performed with tiles 19HBA and 19HCA with both models. The result with XGBoost returns the complete tiles, making the process less optimal. On the contrary, the result with Random Forest returns the tiles cut according to the region.

Associated sits API function Please verify the 'roi' parameter of the sits_classify function

M3nin0 commented 1 week ago

Dear @EstefaPizarro,

Thanks a lot for reporting this issue and for taking the time to provide this detailed description.

We will investigate this issue and return to you as soon as possible.

Cheers, Felipe Carlos