A hybrid approach based on the combination of variable selection using decision trees and case-based reasoning using the Mahalanobis distance: For bankruptcy prediction.
Sungbin ChoHyojung HongByoung-Chun HaPublished in: Expert Syst. Appl. (2010)
Keyphrases
- variable selection
- mahalanobis distance
- bankruptcy prediction
- decision trees
- high dimensional
- cross validation
- euclidean distance
- noise model
- covariance matrix
- model selection
- learning vector quantization
- dimension reduction
- financial data
- machine learning
- data distribution
- literature review
- metric learning
- distance metric
- machine learning algorithms
- high dimensional data
- information gain
- noise level
- unsupervised learning
- dimensionality reduction
- feature space
- objective function
- feature selection