fan Industrial is a machine component that used to expel air in the room to exit the room and maintain air circulation in the room. One component of the fan that often damaged is the bearing. The method commonly used to detect the condition of a bearing is the vibration based method. The method spectrum is one of the vibration-based methods but this method has a disadvantage that is not easily understood by operators in the field. The pattern recognition method is an easy method to use because it does not need to translate graphs spectrum. The pattern recognition method used in this study is Support Vector Machine (SVM). The purpose of this study is to detect fault in inner race in ball bearings. This study uses two different bearing conditions, namely normal bearings and defective bearings. Fault on the bearing are made by the method of Electrical Discharge Machine (EDM) on the inner track with a depth of 1.4 mm and width of 0.4 mm. The test is carried out on an fan test rig industrial using the software MATLAB and recording 700 data files of for each condition. The recorded data was extracted to 17 statistical parameters such as: Standard Deviation (SD), Root Mean Square (RMS), Peak Value, Kurtosis, Crest Factor, Variance, Mean, Entropy, Minimum Value, Standard Error (SE), Skewness, Maximum Value, Range, Sum, Median, Signal to Noise and Distortion Ratio (SINAD), and Signal to Noise Ratio (SNR) ones then visually selected. SVM classification is done with variations kernel Radial Basis Function (RBF), Polynomial and Linear. The results showed statistical parameters Entropy with Standard Error using variations kernel Radial Basis Function (RBF), Polynomial and Linear are recommendations for the classification of fault in inner race bearings because it has an accuracy of 100%.