Pasaribu, Surya Rosauli (2023) Implementasi Case-Based Reasoning untuk Diagnosis Awal Penyakit Anjing Menggunakan Similaritas Jaccard dan Sorensen-Dice Coefficient. Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text
19081010122_Cover.pdf Download (743kB) | Preview |
|
|
Text
19081010122_Bab1.pdf Download (26kB) | Preview |
|
![[img]](http://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text
19081010122_Bab2.pdf Restricted to Repository staff only Download (279kB) | Request a copy |
|
|
Text
19081010122_Bab3.pdf Restricted to Repository staff only Download (1MB) | Request a copy |
|
|
Text
19081010122_Bab4.pdf Restricted to Repository staff only Download (2MB) | Request a copy |
|
|
Text
19081010122_Bab5.pdf Download (12kB) | Preview |
|
|
Text
19081010122_Daftar_Pustaka.pdf Download (83kB) | Preview |
|
|
Text
19081010122_Lampiran.pdf Restricted to Repository staff only Download (399kB) | Request a copy |
Abstract
Dogs are one of the animals that most people keep as pets. However, the owner's lack of knowledge regarding pet diseases can lead to errors in care, even death. Current technological advances make it possible to have a system that can diagnose dog diseases based on the signs/symptoms that appear. This research builds a web-based diagnosis system that applies the Case-Based Reasoning reasoning method as well as the Jaccard similarity and Sorensen-Dice coefficient to find the level of similarity between two objects, namely new cases and old cases. These two similarities are used to see differences in providing solutions for new cases and to determine system accuracy. Testing with 5-fold cross-validation with each fold having 16 test data and 64 training data produces a Confusion Matrix which better describes system performance compared to testing with a 7:3 ratio using 24 test data and 56 training data. By testing 5-fold cross-validation using a threshold of 0.65, Jaccard similarity produces an accuracy of 48.75%, precision of 56.36%, and recall of 50.30%. Meanwhile, Sorensen-Dice similarity produces an accuracy of 88.75%, precision of 92.73%, and recall of 93.64%. When compared with the use of another threshold, namely 0.5, Jaccard similarity produces an accuracy of 88.75%, precision of 92.73%, and recall of 93.64%. Meanwhile, Sorensen-Dice similarity produces an accuracy of 93.75%, precision of 96.36%, and recall of 100%. In Jaccard similarity, there are several disease classes that cannot be identified because they provide similarity values that are below the chosen threshold so that the system will count them as False Negative and have an impact on the Confusion Matrix results. Of course, the Confusion Matrix results are also influenced by the quality of the data set itself. Finally, based on the threshold and Confusion Matrix results in this diagnosis system, the use of Sorensen-Dice similarity is more recommended than Jaccard similarity because the system is more robust in classifying a test case with a different threshold.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | T Technology > T Technology (General) | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Surya Rosauli Pasaribu | ||||||||||||
| Date Deposited: | 24 Nov 2023 09:08 | ||||||||||||
| Last Modified: | 24 Nov 2023 09:08 | ||||||||||||
| URI: | http://repository.upnjatim.ac.id/id/eprint/18804 |
Actions (login required)
 |
View Item |