holisticai.bias.mitigation.DebiasingLearningMF#
- class holisticai.bias.mitigation.DebiasingLearningMF(*args, **kargs)[source]#
Debiasing Learning Matrix Factorization
Debiasing Learning Matrix Factorization handles selection biases by adapting models and estimation techniques from causal inference. The strategy leads to unbiased performance estimators despite biased data, and to a matrix factorization method that provides substantially improved prediction performance on real-world data.
Parameters
- Kint
Specifies the number of dimensions.
- normalizationstr
- Strategy to normalize rating matrix. Avaiables are:
‘Vanilla’,
‘SelfNormalized’
‘UserNormalized’
‘ItemNormalized’
- lamdafloat
Model parameter.
- metric: str
Metric used as cost function.
- clip_val: float
Propensity Clip Value
- seed: int
Random Seed
- bias_mode: str
Bias value using in the model: - “None”: No bias - “Free”: Use bias wihtout regularizer in the cost function. - “Regularized”: Use bias with regularizer in the cost function.
- verboseint
If >0, will show progress percentage.
Examples
>>> from holisticai.bias.mitigation import DebiasingLearningMF >>> mitigator = DebiasingLearningMF(**params) >>> mitigator.fit(data_matrix)
References