Multispectral Land Cover Classification of Wageningen using Sentinel-2 Images
The workflow of the project includes initial data analysis, followed by data preprocessing and feature engineering techniques such as handling class imbalance and adding two additional features. Separate training and validation sets were used, and two different models—K-Nearest Neighbors (KNN) and Random Forest—were implemented, along with hyperparameter tuning to optimize performance. Model evaluation was performed using classification metrics such as accuracy_score, f1_score and confusion_matrix providing insights into each model's performance on the validation set.