Tree Regression (California Housing)

Experiment initialization and data preparation

from piml import Experiment
from piml.models import TreeRegressor

exp = Experiment()
exp.data_loader(data="CaliforniaHousing_trim2", silent=True)
exp.data_prepare(target="MedHouseVal", task_type="regression", silent=True)

Train Model

exp.model_train(model=TreeRegressor(max_depth=6), name="Tree")

Evaluate predictive performance

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

          MSE     MAE       R2

Train  0.0184  0.0979   0.6762
Test   0.0212  0.1059   0.6178
Gap    0.0028  0.0080  -0.0584

Global interpretation starting from the root node

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

Global interpretation starting from the second node

exp.model_interpret(model="Tree", show="tree_global", root=2, depth=3,
                    original_scale=True, figsize=(16, 10))

Local interpretation

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

Estimated memory usage: 27 MB