Leandro L. Minku - Open Source Code and Other Data


Source Code

Incremental and Offline Negative Correlation Learning (Java)

Concept Drift Data Set Generator (Matlab)


 

Data Sets

Artificial Concept Drift Data Sets

Semi-artificial Concept Drift Data Sets

Preprocessed PROMISE holdout sets used in TOSEM 2013 and PROMISE 2013b.

Preprocessed PROMISE data sets used in IST 2013 and PROMISE 2011.


Result Files

Note that the results of different approaches are usually not directly comparable to these results when using different types of base learners.

Diversity for Dealing with Drifts (DDD) -- prequential accuracy and standard deviation for artificial data sets, published in TKDE 2012.

Diversity for Dealing with Drifts (DDD) -- prequential accuracy and standard deviation for real world data sets, published in TKDE 2012.

Obs.: Sev 1, 2 and 3 are for low, medium and high severity. Sp 1, 2 and 3 are for high, medium and low speed.