(Note that the data sets are different for the two tasks.)

E-commerce Transaction Anomaly Classification (Easy)

The following sets of data are availabe for the E-commerce Transaction Anomaly Classification (Easy). They are .zip files, so use Winzip on Windows or unzip (not gunzip) on Linux.

Data Set (.zip files)	File Size	Number of Examples	Summary
Training Data	1 MB	94,682	One line per example, feature values are comma delimited. First line has feature names.
Training Labels	3 KB	94,682	One line per example, parallel to the training data file. 1 means the corresponding training example is positive, 0 means the corresponding training example is negative.
Test Data	1 MB	36,019	Same format as training data.

E-commerce Transaction Anomaly Classification (Hard)

The following sets of data are availabe for the E-commerce Transaction Anomaly Classification (Hard). They are .zip files, so use Winzip on Windows or unzip (not gunzip) on Linux.

Data Set (.zip files)	File Size	Number of Examples	Summary
Training Data	3 MB	100,000	One line per example, feature values are comma delimited. First line has feature names.
Training Labels	5 KB	100,000	One line per example, parallel to the training data file. 1 means the corresponding training example is positive, 0 means the corresponding training example is negative.
Test Data	1 MB	50,000	Same format as training data.