Skip to main content

Table 3 Attribute selector

From: A genetic algorithm based framework for software effort prediction

AS Description
Genetic Search (GS) Uses a simple genetic algorithm. Parameters include population size, number of generations, and probabilities of crossover and mutation.
BestFirst (BF) This attribute selector performs greedy hill climbing with backtracking. It is possible to specify how many consecutive non-improving nodes must be encountered before the system backtracks.
LinearForwardSelection (LFS) This attribute selector is an extension of BestFirst that considers a restricted number of the remaining attributes when expanding the current point in the search.
Backward Elimination (BE) This attribute selector starts with the whole set of attributes and eliminates one attribute in each iteration until no single attribute elimination improves the evaluation of the subset.
Forward Selection (FS) This attribute selector starts from an empty set and evaluates each attribute individually to find the best single attribute. It then tries each of the remaining attributes in conjunction with the best pair to find the best group of the three attributes.