Feature selection is crucial for improving prediction accuracy and reducing training time. This paper introduces SHAP-IEL which combines a SHAP-based feature selection as a filter approach in classification datasets with an Incremental Ensemble Learning (IEL). The approach leverages SHapley Additive exPlanations (SHAP) to rank features initially. Next, the top 10 features are then selected and used in an ensemble learning process, where majority voting is employed to make final predictions. We demonstrated the superiority of SHAP-IEL using three datasets: credit card and bank marketing, representing the marketing and business domain, and SRBCT, a gene expression dataset from the medical domain. The code and datasets are available on the GitHub page at https://github.com/hnematzadeh/SHAP-IEL .