Modeva 2.1.0

Integrated tool for model development and validation.

MoDeVa

Model Development and Validation — an integrated toolkit for building and validating interpretable machine-learning models.

MoDeVa provides a single, consistent workflow for loading data, training a range of interpretable and black-box models, and running a comprehensive test suite for explanation, diagnostics, robustness, fairness and model comparison.

• Documentation: https://modeva.ai
• License: proprietary (offline signed-license activation; a free developer tier is available)

Installation

pip install modeva

That's it — every dependency (numpy, scikit-learn, XGBoost, LightGBM, PyTorch, SHAP, MLflow, …) is resolved and installed automatically. MoDeVa requires Python 3.12.

Verify your environment

If an import fails or a dependency looks off, run the built-in diagnostic. It reads MoDeVa's declared dependencies, checks each against what's installed, and prints the exact pip install line to fix any gap (it never installs anything itself):

import modeva
modeva.check_dependencies()

Quickstart

from modeva import DataSet, TestSuite
from modeva.models import MoXGBRegressor, MoLGBMRegressor

# 1. Load a dataset and make a train/test split
ds = DataSet()
ds.load(name="BikeSharing")
ds.set_random_split()

# 2. Train one or more models
model1 = MoXGBRegressor(name="XGB")
model1.fit(ds.train_x, ds.train_y)

model2 = MoLGBMRegressor(name="LGBM", max_depth=2, verbose=-1)
model2.fit(ds.train_x, ds.train_y)

# 3. Explain and diagnose
ts = TestSuite(ds, model1)
results = ts.explain_pdp("hr")
results.plot()

# 4. Compare models
tsc = TestSuite(ds, models=[model1, model2])
results = tsc.compare_slicing_accuracy(features="hr", method="quantile",
                                       bins=10, metric="MSE")
results.plot()

See the documentation for the full user guide, API reference and notebook examples.

License activation

MoDeVa uses offline, signed-license activation — there is no license server and no machine binding. A free developer tier is available; enterprise tiers unlock the full capability set. See the documentation for how to obtain and activate a license key.