Feature Selection for Very Short-Term Heavy Rainfall Prediction Using Evolutionary Computation

Table 5

Parameters in GA/DE.

GA parameters
Fitness function	𝑘-NN (𝑘=3), 𝑘-VNN (𝑘=3), SVM (type: C_SVC, kernel function: polynomial, linear, and precomputed) [27]
Encoding	Binary (72 dimensions)
No. of populations	20
No. of generations	100
Selection	Tournament selection
Crossover	Multipoint crossover (3 points)
Mutation	Genewise mutation (𝑃=0.005)
Replacement	If an offspring is superior to the worst individual in the population, we replace it with the worst one
DE parameters
Fitness function	𝑘-NN (𝑘=3), 𝑘-VNN (𝑘=3), SVM (type: C_SVC, kernel function: polynomial)
Encoding	Real number (23 dimensions)
No. of populations	20
No. of generations	100
Crossover rate	0.03
FVal	0.05
Replacement	If an offspring is superior to the parent in the population, we replace it with the parent