Data: Consists of traning and testing data. The training set consists of 325 instances of multivariate remote sensing data of some forest in Japan. The training data consists of 28 columns. First column consists of class label, followed by data on 9 bands of spectrum. Last 18 columns consists of predicted minus observed value of every band.
Used hyper parameter tuning for SGD and SVM (kernels: RBF and Poly) models
Selecting model that best explains depedent variable using F1 score