The tremendous developments in technology that have been realized in this digital era have greatly improved the way in which data is collected and used in schools. Over the years the number of secondary schools using technology in processing student data has been increasing steadily. As a result, a large amount of data in electronic form has been gathered. Classification algorithms can be used to study the patterns presented in these data and use it to predict a suitable career for a student. In this study classification algorithms were used to predict a suitable career for form four students. The study evaluated the best classification algorithm for implementing the career recommendation system in Kenya. The Cross Industry Standard Process for Data Mining framework was applied to a dataset drawn from form four students in Bungoma County in Kenya. Stratified random sampling was used to select 50 secondary schools and a 10% of candidates were selected from every sampled schools. The collected data were cleansed, preprocessed and analyzed using a data mining tool of RapidMiner. Various classification algorithms were evaluated in predicting a suitable career for a student. The study findings revealed that classification algorithms can be used to predict a suitable career for a student. All the classifiers that were used gave a predictive accuracy of above 88% though deep learning was the most accurate with 97.5%. However, since the classifiers out performed each other in various metrics, therefore using multiple classification algorithms in building the recommendation model can yield better results. The study therefore concludes that classification models comprising of multiple classifiers can be used to predict suitable careers for secondary students.