Table 1 Overview of the hyperparameter search space and optimal parameters identified using GridSearchCV.

From: A triple pronged approach for ulcerative colitis severity classification using multimodal, meta, and transformer based learning

Algorithm

Hyperparameter space

Best hyperparameters

KNN

n_neighbors: 3, 5, 7, 9

n_neighbors: 9

 

Metric: Euclidean, Manhattan, Minkowski

Metric: Manhattan

SVM

C: 0.1, 1, 10, 100

C: 10

 

Kernel: linear, rbf

Kernel: rbf

 

Gamma: scale, auto

Gamma: scale

Random forest

n_estimators: 100, 200, 300, 400, 500, 600

n_estimators: 600

 

max_depth: none, 10, 20, 30, 40, 50, 60

max_depth: 20

 

min_samples_split: 2, 5, 10

min_samples_split: 5

Back to article page