Diverse CF

---

source

DIVERSECFCONFIG

CLASS relax.methods.diverse.DiverseCFConfig (n_cfs=5, n_steps=1000, lr=0.01, lambda_=0.01, seed=42)

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

---

source

DIVERSECF

CLASS relax.methods.diverse.DiverseCF (configs=None)

Base CF Explanation Module.

from relax.data import load_data
from relax.module import PredictiveTrainingModule, PredictiveTrainingModuleConfigs, load_pred_model
from relax.evaluate import generate_cf_explanations, benchmark_cfs
from relax.trainer import train_model

Load data:

dm = load_data('adult', data_configs=dict(sample_frac=0.1))

/home/birk/miniconda3/envs/nbdev2/lib/python3.8/site-packages/sklearn/preprocessing/_encoders.py:868: FutureWarning: `sparse` was renamed to `sparse_output` in version 1.2 and will be removed in 1.4. `sparse_output` is ignored unless you leave `sparse` to its default value.
  warnings.warn(

Train predictive model:

# load model
params, training_module = load_pred_model('adult')

# predict function
pred_fn = training_module.pred_fn

Define DiverseCF:

diversecf = DiverseCF()

Generate explanations:

cf_exp = generate_cf_explanations(
    diversecf, dm, pred_fn=pred_fn, 
    pred_fn_args=dict(
        params=params, rng_key=random.PRNGKey(0)
    )
)

Evaluate explanations:

benchmark_cfs([cf_exp])

		acc	validity	proximity
adult	DiverseCF	0.8241	0.393932	1.913267