<![CDATA[This paper aims to apply the weight selection feature by considering the Gain Ratio value in the decision tree algorithm in classifying student academic scores. We determine the feature selection from the gain ratio based on the split value information to reduce the feature's (attribute) bias value. The highest Gain Ratio' value will be the root of the branching in the tree in which becomes a determining feature (attribute) of student graduation. We use 82 data which are divide into two classes called a pass and a not pass. From the data, we know that the attribute ip smt 7 got the highest gain ratio value with 0.581. On the other hand, the multimedia introduction attribute got the lowest gain ratio value with 0.070. The calculation model using cross-validation with a value of k = 5 resulted in optimal performance. The resulting accuracy is 79.19% and AUC 0.778 using the decision tree algorithm. The threshold value of the gain ratio used is 1.00 so that four attributes are not used in this paper. feature selection using weights with information gain ratio will select the attribute selection process to be built in the model.]]>
Copyrights © 2023
