A Multi-Criteria Decision-Making Approach for Targeted Distribution of Smart Indonesia Card (KIP) Scholarships
Education is fundamental to developing quality human resources, as stated in the 1945 Constitution. To support this goal, the Indonesian government introduced the Smart Indonesia Program through the KIP scholarship for students from underprivileged families. Recipient selection also involves various...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | en en |
| Published: |
2025
|
| Online Access: | http://ur.aeu.edu.my/1419/1/Thesis%20Komang%20Aryasa.pdf http://ur.aeu.edu.my/1419/2/Thesis%20Komang%20Aryasa-1-24.pdf http://ur.aeu.edu.my/1419/ https://online.fliphtml5.com/sppgg/vvwi/?1768379583279 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Education is fundamental to developing quality human resources, as stated in the 1945 Constitution. To support this goal, the Indonesian government introduced the Smart Indonesia Program through the KIP scholarship for students from underprivileged families. Recipient selection also involves various factors, including poverty indicators, social conditions, and academic performance. This study aims to develop a comprehensive decision-making model for KIP scholarship selection through four main stages. First, the poverty criteria were weighted using three approaches: the Analytical Hierarchy Process (AHP), Entropy, and the hybrid method, followed by a ranking process using the VIKOR method. Second, the clustering process was conducted to group the priorities of prospective scholarship recipients using the K-Means and K-Medoids methods, as well as a combination of PCA+K-Means and PCA+K-Medoids. Third, the classification of scholarship recipient eligibility was performed by comparing the C5.0 and K-Nearest Neighbors (KNN) algorithms. Fourth, the classification results were validated to ensure the accuracy and precision of the decision. The study found that the hybrid weighting model with λ = 0.8 (80% subjective and 20% objective) achieved a ranking stability of 61%, indicating improved accuracy and consistency in selecting KIP scholarship recipients. Sensitivity analysis showed that Hybrid+VIKOR had the lowest change (1.20%) compared to AHP+VIKOR (5.06%) and Entropy+VIKOR (53.71%), confirming its superior stability against weight variations. In the clustering stage, the combination of PCA+KMedoids with two initial medoids produced stable clusters in all iterations, suggesting that K-Medoids provided a better representation of data variation. Meanwhile, in the classification stage, the C5.0 algorithm achieved the highest accuracy of 97.27% from a total of 551 data points, with 80% used as training data and 20% as testing data. This study can be utilised to significantly improve decision-making by introducing opportunities for the development of stronger scientific methodologies and contributions, as well as broader practical relevance, especially in supporting transparent, fair, and data-driven scholarship selection processes. Moreover, the developed approach also had the potential to be applied in various other social policy contexts. |
|---|