Optimized Random Forest Classifier for Predicting Ideal ‎Candidate for The General Election

Electoral systems and candidate selections are the two important pillars of modern democratic elections. The integrity and ‎competence of candidates are significant contributing factors to government enactment. Hence, the selection of competent candidates ‎is an indispensable process in the electoral system. This research proposes a new model, named RANWIN (Random forest ‎classifier-based model for selecting Winning candidates), for predicting the winning probability of the candidate and the party. Our ‎model integrates Random Forest Classifier (RFC) and Bayesian Optimization Technique (BOT) to achieve outstanding ‎performance. It considers several factors for predicting the winning probability of candidates, including the popularity of the candidate, ‎past election history, party change or personalization, vote bank, etc. For predicting the winning probability of the party, this method ‎considers the manifestoes, vote bank, leaders, electoral history, intraparty struggles, major rallies hosted in the constituency, and ‎the strength of alliance parties etc. After calculating the impact of all these parameters, RANWIN uses RFC to predict the winning ‎possibility of the candidates as well as parties. To enhance the efficiency of the intended framework, we apply BOT in the prevailing ‎RFC to find out the optimal hyperparameters of the classification process. To prove the better performance of our framework, its ‎enactment is related to other existing approaches regarding accuracy, precision, the Area Under the receiver operating Characteristic ‎‎(AUC) curve, recall, F-measure, and Root Mean Square Error (RMSE). The empirical results demonstrate that RANWIN can be ‎considered as a more effective method with higher predictive accuracy, precision, AUC, recall, F-measure, and lower RMSE of ‎‎94.32%, 95.5%, 90.6%, 97.6%, 98.0%, and 16.23%, correspondingly. As a result, we can prove that RANWIN is an improved ‎framework for candidate selection and has a positive impact on the current political landscape as related to approaches based on ‎other machine learning techniques. The key goal of this research is to help political parties select the right candidates to win public ‎office‎.