Herdianto, Ivan (2026) ESTIMATION OF LAND AREA BASED ON-SCREEN DIGITIZATION UTILIZING KARNEY'S ALGORITHM AND SUPPORT VECTOR REGRESSION. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
22081010047-cover.pdf Download (1MB) |
|
Text (Bab 1)
22081010047-bab1.pdf Download (157kB) |
|
Text (Bab 2)
22081010047-bab2.pdf Restricted to Repository staff only until 2 September 2029. Download (287kB) |
|
Text (Bab 3)
22081010047-bab3.pdf Restricted to Repository staff only until 2 September 2029. Download (737kB) |
|
Text (Bab 4)
22081010047-bab4.pdf Restricted to Repository staff only until 2029. Download (553kB) |
|
Text (Bab 5)
22081010047-bab5.pdf Download (133kB) |
|
Text (daftar-pustaka)
22081010047-daftarpustaka.pdf Download (175kB) |
|
Text (lampiran)
22081010047-lampiran.pdf Restricted to Repository staff only until 2 September 2028. Download (174kB) |
Abstract
This research examines spatial data accuracy issues in on-screen digitizing, which is susceptible to human error and planar projection distortions. The objective is to compare the accuracy of the planar method with Karney's Algorithm (based on the WGS84 ellipsoid), test the effectiveness of Support Vector Regression (SVR) as a digitizing bias corrector, and assess its consistency across various land area stratum. The study was conducted using 400 land polygon samples in Surabaya, categorized into small, medium, and large land areas, utilizing data from the ATR/BPN BHUMI WebGIS as the official reference. The SVR model was then optimized using GridSearchCV and evaluated through MAPE, R², MAE, RMSE metrics, as well as the Wilcoxon test. Geodesic computation evaluated across all 400 samples demonstrates that Karney's Algorithm is more accurate, achieving a MAPE of 6.89% compared to the planar method (8.41%), with a mean absolute difference of 45.69 m² between the two methods. The application of SVR significantly improved the overall accuracy, achieving a MAPE of 5.34% and an R² of 0.9997. In cross-stratum testing, SVR also demonstrated the best consistency, outperforming both Random Forest Regression and Ridge Regression models. It can be concluded that the integration of Karney's Algorithm and SVR yields an accurate, efficient, and adaptive method for estimating digitized land areas across varying geometric characteristics of land parcels.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | G Geography. Anthropology. Recreation > G Geography (General) Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76.6 Computer Programming |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Ivan Ivan Herdianto Herdianto | ||||||||||||
| Date Deposited: | 02 Sep 2026 08:32 | ||||||||||||
| Last Modified: | 02 Sep 2026 08:42 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/59834 |
Actions (login required)
 |
View Item |