Comparing the Prediction of Numeric Patterns on Form C1 Using the K-Nearest Neighbors (K-NN) Method and a Combination of K-Nearest Neighbors (K-NN) with Connected Component Labeling (CCL)

Abstract

Indonesia's elections serve as a cornerstone of its democratic system, with the active participation of its citizens being of paramount importance. To bolster transparency and civic engagement during these elections, the SITUNG system (Election Result Information System) is employed for the tabulation of election results. However, the current tabulation process remains manual, potentially leading to data entry errors and a reduced accuracy of election outcomes. This research endeavor seeks to enhance the efficiency and accuracy of election result tabulation by employing the K-Nearest Neighbors (K-NN) method for recognizing numeric patterns on Form C1, both independently and in combination with Connected Component Labeling (CCL). The K-NN method demonstrates a commendable 60.0% accuracy in recognizing numeric patterns from the original Form C1 data. However, when combined with CCL, the accuracy drops to 51.2%. This research makes a significant contribution by simplifying the tabulation process and improving the accuracy of election results in Indonesia through the application of the K-NN method. The technology is anticipated to fortify democracy by promoting a more transparent and participatory electoral process for the citizens.

Keywords

• predicting numeric patterns
• k-nearest neighbors method (k-nn)
• connected component labeling (ccl)
• general election commission form c1