|
|
 |
| |
|
|
|
Point-Biserial Correlation Analysis of Fuzzy Attributes |
|
|
|
PP: 439S-444S |
|
|
|
Author(s) |
|
|
|
Hao-En Chueh,
|
|
|
|
Abstract |
|
|
| Previous studies have presented correlation analyses among fuzzy attributes. This study proposes a correlation analysis method among crisp attributes and fuzzy attributes to investigate their linear relationships. If A and B are two attributes, A is a nominal-dichotomous attribute, and B is a fuzzy attribute. Thus, identifying the relationship between A and B is possible by using the point-biserial correlation analysis. The fuzzy point-biserial correlation analysis is defined and derived, using the membership grades of fuzzy attributes. Using the fuzzy point-biserial correlation analysis is feasible if A is a bimodal distribution, but not a real nominal-dichotomous attribute, such as crime youth versus non-crime youth, illiterate versus non-illiterate person, and mental deficiency versus normal intelligence. An experimental sample illustrates that our proposed method is simpler and faster than the traditional correlation analysis. |
|
|
|
|
|
|
|
