DATA MINING UNTUK PREDIKSI KEBANGKRUTAN PERUSAHAAN BERDASARKAN DATA KUALITATIF

Qualitative data are subjective and usually more difficult to measure. The objective of this research is predicting qualitative bankruptcy using data mining with neural network model. Multilayer Perceptron (MLP) and Radial Basis Function (RBF) as neural network algorithms are chosen to compare because these algorithms are commonly used. The data taken from UCI machine learning repository with 250 records and 6 attributes and SPSS Neural Network 17.0 is used to implement it. The Confusion Matrix (CM) and ROC (Receiver Operating Characteristic) Curve method is used to measure the performance of the both algorithms. Based on the test results of the implementation, proved that MLP algorithm has higher value of accuracy than RBF algorithm. Using Confusion Matrix, MLP algorithm has higher value of accuracy with 98.7% than RBF algorithm with 97.4%. Using the ROC Curve, MLP algorithm also has higher AUC (Area Under the Curve) value with 1.000 than the RBF algorithm with 0.998. From both of the algorithms, the accuracy values are included as exellent classification, because the AUC value is in the range of 0.90 until 1.00.