Analisis Diskriminan Fisher Menggunakan Metode Two-Fold Cross Validation (Studi Kasus: Klasifikasi Status Kredit Nasabah Di PT Bank Sinarmas)
| Main Author: | Suletandung, Kridayantri S. |
|---|---|
| Format: | Article |
| Terbitan: |
, 2018
|
| Subjects: | |
| Online Access: |
http://repository.unhas.ac.id/handle/123456789/27097 |
Daftar Isi:
- Fisher discriminant analysis is a multivariate analysis used to find the basis of classification of an individual into groups that have been determined, based on the calculation of discriminant score. Fisher discriminant analysis does not require the assumption of normal multivariate distribution, but in its application the homogeneity assumption of the covariance matrix must be satisfied. This study discusses discriminant scores approximation on the Fisher discriminant model using the Two-Fold Cross Validation method. The Fisher discriminant analysis model generated for the first experiment was: Y = 1.7079 x_1 + 0.0783 x_2- 0,0176 x_3 - 0,0708 x_4- 1,9592 x_5- 2,0995 x_6, while for the second experiment obtained discriminant model Fisher: Y = 2,3412 x_1+ 0,5325 x_2 - 0,6625 x_3- 2,3521 x_4 - 0,9169 x_5 -2,4276 x_6. Based on the analysis, the second experimental classification result is better than the first experiment. Therefore, the best model that can be used in the classification of credit status of customers in PT Bank Sinarmas South Veteran Branch are: Y = 2,3412 x_1+ 0,5325 x_2 - 0,6625 x_3- 2,3521 x_4 - 0, 9169 x_5 -2,4276 x_6. Based on the model it is known that every increase of one unit x_1 and x_2 will give positive effect to discriminant score of 2,3412 and 0.5325 respectively, whereas each increase of one unit x_3, x_4, x_5 and x_6 will give negative effect to discriminant score of 0.6625 , 2.3521.0.9169, and 2.4276. Key Words: Fisher discriminant analysis, Two-Fold Cross Validation, Classification