Sharing pattern submodels for prediction with missing values
This addresses the challenge of handling missing values in machine learning applications, particularly when missingness depends on unobserved factors, offering a solution that balances specialization and information sharing.
The paper tackles the problem of making predictions with missing values by proposing sharing pattern submodels, which achieve robust predictions, maintain or improve predictive power, and offer interpretability through a short description, as demonstrated in classification and regression experiments on synthetic and real-world datasets.
Missing values are unavoidable in many applications of machine learning and present challenges both during training and at test time. When variables are missing in recurring patterns, fitting separate pattern submodels have been proposed as a solution. However, fitting models independently does not make efficient use of all available data. Conversely, fitting a single shared model to the full data set relies on imputation which often leads to biased results when missingness depends on unobserved factors. We propose an alternative approach, called sharing pattern submodels, which i) makes predictions that are robust to missing values at test time, ii) maintains or improves the predictive power of pattern submodels, and iii) has a short description, enabling improved interpretability. Parameter sharing is enforced through sparsity-inducing regularization which we prove leads to consistent estimation. Finally, we give conditions for when a sharing model is optimal, even when both missingness and the target outcome depend on unobserved variables. Classification and regression experiments on synthetic and real-world data sets demonstrate that our models achieve a favorable tradeoff between pattern specialization and information sharing.