Determination of accuracy value in id3 algorithm with gini index and gain ratio with minimum size for split, minimum leaf size, and minimum gain

Abstract

A process that explains and functions to distinguish data classes is called Classification. The use of Gain Ratio in ID3 algorithm is very influential on accuracy compared to the Gini Index, and if the higher the determination of the minimum size of split, the minimum leaf size and the minimum gain, the accuracy results will be greater in the gini index and gain ratio, but from both methods there is an accuracy difference of 27% for each accuracy test. In determining the minimum size of split 2, the minimum leaf size 2 and the minimum gain 0.4 with the value of accuracy at the gain ratio of 85.53% and gini index 58.67%. The minimum size of split 6, the minimum leaf size 3 and the minimum gain 0.8 with accuracy values at gain ratio 64.67% and gini index 60.67%. Then the minimum size of split 12, the minimum leaf size 6 and the minimum gain of 0.16 with the value of accuracy at the gain ratio of 86.00% and the index of 68.00%. While the minimum size of split 48, the minimum leaf size 24 and the minimum gain of 0.64 with the value of accuracy at the gain ratio of 95.33% and the index value of 72.00%. Then the Gain ratio which produces the highest accuracy value compared to the gini index.