Missing Values Imputation

Weka package for missing values imputation (and injection) using various techniques.

The following two filters are available:

weka.filters.unsupervised.attribute.MissingValuesImputation - for imputing missing values
weka.filters.unsupervised.attribute.MissingValuesInjection - for injecting missing values

Imputation

The imputation techniques listed below are available through the weka.filters.unsupervised.attribute.MissingValuesImputation filter:

NullImputation - dummy
MeansAndModes - like WEKA's ReplaceMissingValues filter
MultiImputation - applies the specified imputation algorithms sequentially
SimpleNearestNeighbor - uses nearest neighbor approach to determine most common label or average (date/numeric)
SupervisedPrediction - predicts missing values in a range of attributes by using regression/classification algorithms built on this attribute subset with the attribute that gets imputed as class attribute and the remainder of the attributes as input variables.
UserSuppliedValues - simply replaces missing values with user-supplied ones
IRMI - M. Templ et al (2011): Iterative stepwise regression imputation using standard and robust methods (contributed by Chris Beckham)

Injection

The injection techniques listed below are available through the weka.filters.unsupervised.attribute.MissingValuesInjection filter:

NullInjection - dummy
MultiInjection - applies the specified injection algorithms sequentially
AllWithinRange - set all specified attributes to missing
ClassOnly - only sets the class values to missing
RandomPercentage - sets random percentage of values in selected attribute range to missing
Regex - replaces strings that match the regular expression in nominal and string attributes
Values - replaces the specified strings in nominal and string attributes

Releases

Click on one of the following links to download the corresponding Weka package:

How to use packages

For more information on how to install the package, see:

https://waikato.github.io/weka-wiki/packages/manager/

Maven

Add the following dependency in your pom.xml to include the package:

    <dependency>
      <groupId>com.github.fracpete</groupId>
      <artifactId>missing-values-imputation-weka-package</artifactId>
      <version>2022.6.29</version>
      <type>jar</type>
      <exclusions>
        <exclusion>
          <groupId>nz.ac.waikato.cms.weka</groupId>
          <artifactId>weka-dev</artifactId>
        </exclusion>
      </exclusions>
    </dependency>

Please note, when using Maven you may have to register the imputation/injection class hierarchies with Weka's GenericObjectEditor if you want to use them in the GUI as well. See the following files:

JohnnyLin12/missing-values-imputation-weka-package