Mereck
McGowan

NN-kNN for Regression: Accurate Prediction from Interpretable Retrieval

Xiaomeng Ye, Yu Wang, David Leake, David Crandall, Great Abhieyighan, Mereck McGowan

Neural Network k-Nearest Neighbor (NN-kNN) was proposed as an interpretable network model that learns feature weights and similarity to retrieve relevant cases for classification. This paper extends it to regression with the goal of generating accurate predictions based on neighboring cases with similar labels. We introduce three modular components: an attention mechanism that weights retrieved cases, a locality-aware regularizer that favors label-similar neighbors, and an optional case adaptation module. Across synthetic and standard tabular regression benchmarks, NN-kNN achieves competitive predictive error against strong baselines while providing interpretable label-similar case explanations and supporting manual knowledge injection through human-comprehensible feature weights.