In　recent　decades,　various　machine　learning　techniques　have　been　applied　to　intrusion　detection,　with　Support　Vector　Machines　(SVM)　being　considered　an　effective　method.　However,　most　research　methods　based　on　SVM　tend　to　neglect　the　importance　of　data　quality,　which　is　crucial　for　developing　high-performance　intrusion　detection　systems.　This　paper　explores　the　effectiveness　of　the　filter　feature　selection　method.　Specifically,　the　Chi-square　test,　Pearson　Correlation　Coefficient,　and　Maximum　Information　Coefficient　are　used　as　evaluation　criteria　for　feature　selection　to　obtain　high-quality　training　data.　Subsequently,　the　selected　data　is　used　to　train　an　SVM　classifier　to　establish　an　intrusion　detection　model.　Experiments　were　conducted　on　two　relatively　new　datasets　in　the　field　of　intrusion　detection,　UNSW-NB15　and　CICIDS2017.　The　experimental　results　indicate　that　the　Pearson　Correlation　Coefficient　and　Maximum　Information　Coefficient　methods　are　more　effective　and　stable.　©　2024　IEEE.