FIGS Classification (Taiwan Credit)

Experiment initialization and data preparation

from piml import Experiment
from piml.models import FIGSClassifier

exp = Experiment()
exp.data_loader(data="TaiwanCredit", silent=True)
exp.data_summary(feature_exclude=["LIMIT_BAL", "SEX", "EDUCATION", "MARRIAGE", "AGE"], silent=True)
exp.data_prepare(target="FlagDefault", task_type="classification", silent=True)

Train Model

exp.model_train(model=FIGSClassifier(max_iter=100, max_depth=4), name="FIGS")

Evaluate predictive performance

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

           ACC      AUC       F1 LogLoss   Brier

Train   0.8246   0.7891   0.4926  0.4205  0.1312
Test    0.8218   0.7637   0.4636  0.4378  0.1357
Gap    -0.0028  -0.0255  -0.0290  0.0173  0.0045

Global interpretation for the splits heatmap

exp.model_interpret(model="FIGS", show="figs_heatmap", tree_idx=0, figsize=(12, 4))

Global interpretation for the first tree

exp.model_interpret(model="FIGS", show="tree_global", tree_idx=0, root=0,
                    depth=3, original_scale=True, figsize=(16, 10))

Global interpretation for the second tree

exp.model_interpret(model="FIGS", show="tree_global", tree_idx=1, root=0,
                    depth=3, original_scale=True, figsize=(16, 10))

Local interpretation for the first tree

exp.model_interpret(model="FIGS", show="tree_local", sample_id=0, tree_idx=0,
                    original_scale=True, figsize=(16, 10))

Local interpretation for the second tree

exp.model_interpret(model="FIGS", show="tree_local", sample_id=0, tree_idx=1,
                    original_scale=True, figsize=(16, 10))

Estimated memory usage: 43 MB