Implementation of Machine Learning Algorithm Using K-Nearest Neighbors Technique to Predict Indonesian State Budget Deficit

The state budget of Indonesia is an instrument used by the government to manage the budget to archive development goals in the economy. Since 1984 to 2019 state budget of Indonesia experiences a budget deficit where state expenditure is bigger than its revenue which has an impact on the decline of the trade balance, a decrease in the level of gross domestic product that indicates the ability of the country's economic resources to weaken, and the increase in government debt to finance the budget deficit. This study applies the machine learning algorithm using the k-Nearest Neighbors classification technique to predict Indonesia's State budget deficit by using the nearest optimum distance of the k-Fold Cross-Validation algorithm. The results showed that the application of the budget deficit prediction can predict a decrease/increase in the budget deficit with an accuracy level of 63%. This level of accuracy is obtained by using the top 9 nearest neighbors distance that is most appropriate for this study