In　this　paper,　we　present　a　consumption　pattern　recognition　system　based　on　SVM.　It　can　produce　an　optimized　classification　pattern　using　SVM　algorithm　and　use　the　pattern　to　predict　consumer　behaviors.　In　this　system,　three　dimension　reduction　methods　including　Principal　Component　Analysis　(PCA),　correlation　analysis　and　data　cubes　are　applied　to　reduce　dimension　of　features　and　two　training　methods　including　Support　Vector　Machine　(SVM)　and　Support　Vector　Machine　by　Increasing　Negative　Examples　(SVM-INE)　are　utilized　to　build　classifiers.　Consumption　pattern　recognition　system　can　find　the　consumption　habits　of　specific　consumer　group　which　are　helpful　to　well-targeted　marketing.　Empirical　results　show　that　the　system　can　recognize　different　consumption　pattern　with　high　efficiency　and　accuracy.　©　2011　IEEE.