|
|
|
|
| Missing data |
Missing feature values in a dataset can be coded as a NaN. Almost any PRTools operation that involves NaN element will result in a NaN as well. For that reason it is necessary to modify such a dataset before further preprocessing. The routine misval offers the following options:
| > Options of misval for handling missing values in datasets | |
| remove | remove objects (rows) that contain missing values (default) |
| f-remove | remove features (columns) that contain missing values |
| mean | fill the entries with the mean of their features |
| c-mean | fill the entries with the class mean of their features |
| <value> | fill the entries with a fixed constant |
R.P.W. Duin
, January 28, 2013|
|
|
|
| Missing data |