In Bangladesh's higher academies, drug and alcohol addiction has recently become a serious risk factor for pupils, especially female students. Substance abuse and mental health problems have been linked to academic difficulties among female undergraduate students, according to correlations observed among graduate students. Students in higher education who are addicted to drugs participate in a range of criminal actions are unable to complete their education, and eventually make an attempt at suicide. As academics, we must therefore take action to shield these impressionable minds from potentially fatal addiction. In this study, machine learning is used to identify the factors that contribute to female drug addiction and forecast when it will occur. First, we interview female university students, both addicted and not. We apply six notable machine learning algorithms—CNN, SVM, RNN, Naive Bayes, Decision Trees, and Ensemble—on the pre-processed data set. Next, we evaluate the performance of each of these classifiers with respect to several important performance indicators. By achieving an accuracy close to 98.0%, Decision Tree is determined to surpass all other classifiers in terms of all criteria.