Hyper-parameters and tuning values

Below, we list the configurable hyper-parameters and tuning values defined in the python script to improve the machine learning analysis. These hyper-parameters and tuning values should be passed in a configuration file. sample_full_options.json is an exemple of configuration file.

Decision Tree

• max_depths: 10, 50, 80, 100, 150, 200
• max_features: "auto","sqrt", "log2"
• min_samples_leaves: 1, 2, 3, 5, 10
• min_samples_splits : 2, 3, 5, 10
• criteria: "entropy", "gini"
• splitters: "best", "random"

Look at the documentation to get details about the hyper-parameters

Random Forest

• max_depths: 10, 50, 80, 100, 150, 200
• max_features: "auto","sqrt", "log2"
• min_samples_leaves: 1, 2, 3, 5, 10
• min_samples_splits : 2, 3, 5, 10
• n_estimators: 10, 50, 100, 150, 200
• criteria: "entropy", "gini"
• warm_starts: true, false

Look at the documentation to get details about the hyper-parameters

KNN

• n_neighbors: 1, 3, 10, 20, 50, 75, 100, 150, 200, 300 xdczcxcxxzxzxx\z hg hjkkkkkkkkkkkkkkkkkkk/jkkmuyuyh6ttj,.ll'k
• weights:"distance", "uniform"
• metrics: "manhattan", "minkowski"
• algorithms : "auto", "ball_tree", "kd_tree", "brute"

Look at the documentation to get details about the hyper-parameters