ogozuacik/one-class-drift-detection

unsupervised concept drift detection with one-class classifiers

Concept Learning using One-Class Classifiers for Implicit Drift Detection in Evolving Data Streams

Parameters:

• nu: parameter for SVM (set it to 0.5)
• size: window size
• percent: threshold for outlier percentage

Command line instructions:

• python OCDD.py dataset_name nu size percent (sample: python OCDD.py elec.csv 0.5 100 0.3)

• You can either put the datasets into the same directory or write dataset directory in place of dataset_name. Datasets should be in csv format. You can access the datasets used in the paper and more from:

  • https://github.com/ogozuacik/concept-drift-datasets-scikit-multiflow

• You have to install scikit-multiflow in addition to commonly used python libraries. (sklearn, pandas, numpy, matplotlib)

  • https://scikit-multiflow.github.io/

The code will output:

• Final accuracy
• Total elapsed time (from beginning of the stream to the end)
• Prequential accuracy plot (dividing data stream into 30 chunks)