Train a model to produce the Land use and land cover map for Estonia using XGboost
The sentinel-2 satellite data was used to train the XGboost model. The performance is as follow:
Training performance of XGB:
0.9113002645447547
precision recall f1-score support
1 0.91 0.97 0.94 2923868
2 0.91 0.94 0.93 1716062
3 0.83 0.37 0.51 319692
4 0.93 0.81 0.87 249149
5 0.95 0.96 0.95 195419
6 0.90 0.58 0.71 75792
7 0.86 0.81 0.84 23135
8 0.57 0.38 0.45 18444
accuracy 0.91 5521561
macro avg 0.86 0.73 0.77 5521561
weighted avg 0.91 0.91 0.90 5521561
________________________________________________________________________________
Test performance of XGB:
0.9068290293876948
precision recall f1-score support
1 0.93 0.97 0.95 865578
2 0.83 0.93 0.88 221242
3 0.82 0.32 0.46 67557
4 0.91 0.66 0.76 54820
5 0.96 0.97 0.96 32110
6 0.90 0.54 0.67 13528
7 0.68 0.71 0.70 1492
8 0.55 0.34 0.42 3486
accuracy 0.91 1259813
macro avg 0.82 0.68 0.73 1259813
weighted avg 0.90 0.91 0.90 1259813