Effect of Imputation Methods in the Classifier Performance
Missingvalues in a data set present an important problem for almost any traditionaland modern statistical method, since most of these methods were developed underthe assumption that the data set was complete. However, in the real world nocomplete datasets are available and the issue of missing data...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Sakarya University
2019-12-01
|
| Series: | Sakarya Üniversitesi Fen Bilimleri Enstitüsü Dergisi |
| Subjects: | |
| Online Access: | https://dergipark.org.tr/tr/download/article-file/828462 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Missingvalues in a data set present an important problem for almost any traditionaland modern statistical method, since most of these methods were developed underthe assumption that the data set was complete. However, in the real world nocomplete datasets are available and the issue of missing data is frequentlyencountered in veterinary field studies as in other fields. While imputation ofmissing data is important in veterinary field studies where data mining isnewly starting to be implemented, another important issue is how it should beimputed. This is because in many studies observations with any variables havingmissing values are being removed or they are completed by traditional methods.In recent years, while alternative approaches are widely available to preventremoval of observations with missing values, they are being used rarely. Theaim of this study is to examine mean, median, nearest neighbors, mice andmissForest methods to impute the simulated missing data which is the randomlyremoved with varying frequencies (5 to 25% by 5%) from original veterinarydataset. Then highly accurate methods selected to impute original dataset forobservation of influence in classifier performance and to determine the optimalimputation method for the original dataset. |
|---|---|
| ISSN: | 2147-835X |