Genetic algorithms for cross-calibration of categorical data

Abstract

The probabilistic problem of cross-calibration of two categorical variables is addressed. A probabilistic forecast of the categorical variables is obtained based on a sample of observed data. This forecast is the output of a genetic algorithm based approach, which makes no assumption on the type of relationship between the two variables and applies a scoring rule to assess the fitness of the chromosomes. It converges to a good-quality point probability forecast of the joint distribution of the two variables. The proposed approach is applied both at stationary points in time and across time. Its performance is enhanced when additional sampled data is included, and can be designed with different scoring rules or made to account for missing data.

Tree-Based Contrast Subspace Mining for Categorical Data

“…In real-life applications, many research works have been conducted to deal with categorical data in various data mining problem [4,5,[8][9][10][11][12][13][14]. This indicates the importance of effectively mining contrast subspace to learn categorical datasets.…”