Analysis of the Ensemble Method Classifier's Performance on Handwritten Arabic Characters Dataset

Arabic character handwriting is one of the patterns and characteristics each person's writing. This characteristic makes writing more challenging if letter recognition process based on a dataset scripts. script has been presented in totaling 16800, representing class hijaiyah letters starting from alif to yes, consisting 600 data for class. The accuracy used can be increased using ensemble method. By multiple algorithms at simultaneously, technique raise level or result score machine learning. study's primary goal evaluate method classifier's performance datasets handwritten characters. classifier uses by applying proposed soft voting provide multiclass classification three learning algorithms, namely, SVM, Random Forest, Decision Tree classification. research produces an value 0.988 several other SVM with 0.103, random forest 1.0, decision tree 0.134. test results confusion matrix evaluation model, including accuracy, precision, recall, f1-score 0.99.