<![CDATA[Customs has the responsibility as a collector of state revenue. Revenue has an important role in supporting infrastructure development. To manage revenue, prediction is needed to make a good APBN planning. To control revenue, predictions are needed as a prerequisite for good planning of the National Budget (APBN). Prediction is used as an action to optimize and control reception. However, revenue prediction are difficult to do because of the revenue influenced by external factors that difficult to predict. Therefore, logical and accountable agreements are needed to to revenue prediction. Predictions are used to prevent actual are lower than predetermined targets thereby increasing revenue that can be controlled because it has an impact on economic growth in Indonesia. The prediction method used is Support Vector Regression (SVR). This algorithm has a strong performance to recognize time series dataset patterns and provides good prediction results if the parameters are well determined because their performance is very dependent on the parameters within them. SVR implementation in this study uses RBF kernel with parameter variation values, namely sigma = 0.13, lambda = 3.29, cLR = 0.02, epsilon = 0.00001 and C = 10, iteration = 15000 and using 4 data features produce the best MAPE <20% so that it can be categorized that SVR is accurate in predicting customs revenue.]]>
Copyrights © 2019
