Obesity poses a significant health risk worldwide, necessitating accurate classification for effective intervention. Hence proposed a novel approach for obesity classification using a Random Forest (RF) classifier optimized by the Binary Altruistic Teaching (BAT) algorithm. The BAT algorithm helps in automatically tuning into regions where promising solutions to the optimization problem may be found, exhibiting a rapid convergence rate. The proposed work consists of seven classes for the obesity classification. The proposed work improved classification accuracy based on BMI, age, gender, and lifestyle factors. Experimental results on benchmark datasets demonstrate the efficacy of our method. The dataset for the proposed work is collected from Kaggle website with 17 features. The Optimised Random Forest Classifier achieves the highest accuracy of 0.9819 as compared to the other classifiers such as LGBM, Decision Tree, Extra Trees, and Random Forest. This advancement could aid healthcare professionals in designing targeted interventions and personalized treatment plans for obesity-related complications.