SVM Parameter Tuning with Grid Search and its Impact on Reduction of Model Over-fitting Petre Lameski 1 ⋆ , Eftim Zdravevski 1 , Riste Mingov 2 , and Andrea Kulakov 1 1 Faculty of Computer Science and Engineering Ss.Cyril and Methodius University, Skopje, Macedonia {petre.lameski,eftim.zdravevski,andrea.kulakov}@finki.ukim.mk 2 NI TEKNA - Intelligent Technologies, Negotino, Macedonia riste.mingov@ni-tekna.com Abstract. In this paper we describe our submission to the IJCRS’15 Data Mining Competition, which is concerned with prediction of dan- gerous concentrations of methane in longwalls of a Polish coalmine. We address the challenge of building robust classification models with sup- port vector machines (SVMs) that are built from time series data. More- over, we investigate the impact of parameter tuning of SVMs with grid search on the classification performance and its effect on preventing over- fitting. Our results show improvements of predictive performance with proper parameter tuning but also improved stability of the classification models even when the test data comes from a different time period and class distribution. By applying the proposed method we were able to build a classification model that predicts unseen test data even better than the training data, thus highlighting the non-over-fitting properties of the model. The submitted solution was about 2% behind the winning solution. Keywords: Support Vector Machines, SVM, Grid Search, Over-fitting, Parameter Tuning, Time Series, Coalminig 1 Introduction In general, mining is associated with work in hazardous conditions. Miners in an underground coalmine can face many threats, such as, methane explosions or rock-burst. The coal mining industry is the leading cause of fatal injuries in the United States [1]. Furthermore, not only accidents but also exposure to lethal gases can lead to long term diseases of miners [2]. According to the National Institute for Occupational Safety and Health, the fatality rate for coal mining in 2006 was 49.5 per 100,000 workers, more than 11 times greater than the fatality rate in all private industry. Inhaling coal dust also causes black lung disease in coal mine workers. Aiming to provide protection for miners, systems for active monitoring of production processes are usually used. One of their fundamental ⋆ This work was partially financed by the Faculty of Computer Science and Engineer- ing at the Ss.Cyril and Methodius University, Skopje, Macedonia.