Interpretation of the Logistic Regression Model Using the SHAP Method in the Classification of Zakat Recipient Eligibility in Sei Rampah Village

Abstract

Accurate zakat distribution requires a recipient determination process that is objective, consistent, and accountable. This study aims to build a Logistic Regression model for classifying zakat recipient eligibility in Sei Rampah Village and to interpret the prediction results using the SHAP (Shapley Additive Explanations) method. The data used in this study is secondary data of zakat recipient candidates obtained from the zakat management committee in Sei Rampah Village, consisting of 1,000 records, with variables including income, number of dependents, occupation, housing condition, and asnaf category. The research stages include data preprocessing, data structure standardization, data cleaning, categorical variable encoding, numerical feature normalization, splitting the dataset into training and testing sets with an 80:20 proportion, building the Logistic Regression model, and evaluating the model using accuracy, precision, recall, and F1-score. The evaluation results show that the model achieved 99% accuracy, 100% precision, 98.28% recall, and 99.13% F1-score.In addition to achieving excellent classification performance, this study also shows that the SHAP method is able to provide transparent interpretations of the model’s decisions, both globally and locally. In the global interpretation, the most influential features in predicting eligibility are income, number of dependents, housing condition, and occupation. In the local interpretation, SHAP is able to explain the contribution of each feature to the prediction for specific individuals. Therefore, the combination of Logistic Regression and SHAP can be used as an effective approach to support the classification of zakat recipient eligibility in an accurate, transparent, and accountable manner.