Regression Conformal Prediction with Nearest Neighbours

Abstract

In this paper we apply Conformal Prediction (CP) to the k-Nearest Neighbours Regression (k-NNR) algorithm and propose ways of extending the typical nonconformity measure used for regression so far. Unlike traditional regression methods which produce point predictions, Conformal Predictors output predictive regions that satisfy a given confidence level. The regions produced by any Conformal Predictor are automatically valid, however their tightness and therefore usefulness depends on the nonconformity measure used by each CP. In effect a nonconformity measure evaluates how strange a given example is compared to a set of other examples based on some traditional machine learning algorithm. We define six novel nonconformity measures based on the k-Nearest Neighbours Regression algorithm and develop the corresponding CPs following both the original (transductive) and the inductive CP approaches. A comparison of the predictive regions produced by our measures with those of the typical regression measure suggests that a major improvement in terms of predictive region tightness is achieved by the new measures

Local Clustering Conformal Predictor for Imbalanced Data Classification

“…CP model has aroused increased interest in the literature of machine learning and has been applied successfully in the classification of medical data [9], image data [10], and so on. Besides classification, CP has been extended to regression [11][12],feature selection [13], and so on. At the aspect of the framework of CP, some modified models have been proposed to improve the flexibility of CP.…”

Distance Metric Learning-Based Conformal Predictor

“…What's more, the CP algorithm is online by nature, so the CP algorithm has great superiority in practical applications [2][3][4]. In previous studies, K nearest neighbors (KNN) classifier was often used to design the nonconformity measure function for CPs, which is known as CP-KNN algorithm [6]. There are some disadvantages in CP-KNN [8].…”

Reliable Probabilistic Prediction for Medical Decision Support

“…To date many different CPs have been developed, see e.g. [14,15,17,18,19,21,22], and have been applied successfully to a variety of important medical problems such as [3,6,9,10,16]. This paper focuses on an extension of the original CP framework, called Venn Prediction (VP), which can be used for making multiprobability predictions.…”