Predicting the travel mode choice is an important task of transportation planning and policy making to understand inter-urban mobility. It enables the enhancement of the third step of the widely used four-step model. While advances in machine learning have led to numerous powerful predictive models, their usefulness for modeling travel mode choice remains none widely explored. The aim of this paper is to fill in this gap by proposing an advanced machine learning approach tailored to this problem. That is, using extensive Moroccan travel diary data in the year 2016, enriched with numerous individual and household features, our contribution consists of investigating the importance of applying the feature selection approach while using support vector machines (SVM) as a predictive model. The experimental results show that the adopted approach outperforms both native SVM and the artificial neural network, which are the most common data-driven techniques of dealing with such a problem.