Over a million people die from lung illness each year worldwide. Since the bulk of the cells that create tumours are enclosed with one another and develop quickly, early diagnosis of lung illness can be challenging. Throughout the course of therapy, tumour detection handling systems, which are often used for the develop a cutting-edge method for lung tumour identification, a diagnosis of lung cancer is essential. Machine learning algorithms for cancer categorization and detection have recently gained popularity and acceptance. The effectiveness of cancer illness prediction using the suggested attribute selection measure is examined in this proposed study utilising a variety of supervised machine learning methods, including the support vector machine, Naive Bayes and Random Forest. Each model's effectiveness is contrasted in in an effort to identify the most effective, optimized algorithm. Experimental findings demonstrate the great computational efficiency of the suggested model in terms of accuracy. The multi-class Decision Tree classifier revealed a higher accuracy of 89.8%, 83%, and 93.1%, and each of the three datasets with 200, 500 and 1000 features extraction respectively.