GAM Classification (CoCircles)

Experiment initialization and data preparation

from piml import Experiment
from piml.models import GAMClassifier

exp = Experiment()
exp.data_loader(data="CoCircles", silent=True)
exp.data_prepare(target="target", task_type="classification", silent=True)

Train Model

exp.model_train(model=GAMClassifier(spline_order=2, n_splines=20, lam=0.6), name="GAM")

Evaluate predictive performance

exp.model_diagnose(model="GAM", show="accuracy_table")

          ACC     AUC      F1  LogLoss    Brier

Train  0.8363  0.9226  0.8387   0.3512   0.1130
Test   0.8475  0.9306  0.8432   0.3406   0.1062
Gap    0.0112  0.0080  0.0045  -0.0105  -0.0068

Global interpretation: effect plot for X0

exp.model_interpret(model="GAM", show="global_effect_plot", uni_feature="X0",
                    original_scale=True, figsize=(5, 4))

Global interpretation: effect plot for X1

exp.model_interpret(model="GAM", show="global_effect_plot", uni_feature="X1",
                    original_scale=True, figsize=(5, 4))

Global interpretation: feature importance

exp.model_interpret(model="GAM", show="global_fi", figsize=(5, 4))

Local interpretation

exp.model_interpret(model="GAM", show="local_fi", sample_id=0, original_scale=True, figsize=(5, 4))

Estimated memory usage: 12 MB