<![CDATA[One of the most commonly used indicators to measure the inflation rate is Consumer Price Index (CPI). Based on the consumer price index metadata published by Bank Indonesia in 2016, housing, water, electricity, gas and fuel group is the CPI group which has the highest proportion of living cost from other CPI groups, which is 25.37 %. In this research, CPI will be predicted by using Support Vector Regression (SVR) method. The stages of the SVR method include normalization of data, calculates Hessian matrix by using Radial Basis Function (RBF) kernel function, sequential learning process, calculate the regression function to get predicted results and evaluates predicted results with Mean Absolute Percentage Error (MAPE). The test results show the minimum MAPE value obtained by 4.271% with the parameter value σ = 50; λ = 1; cLR = 0.0005; ε = 0.0005; C = 1000; the number of training data is 36 for 12 testing data with 100 iterations. The average of predicted results obtained is 112.19605 with the average of the difference between the actual data and the predicted result is 1.52645.]]>
Copyrights © 2018
