Introduction to 2DSyntheticData.zip

This Zip file include a set of 2-D real value data to test performace of anomaly detection algorithm. All points are with [0,1]^2.

There are totally 38 pairs of data. Each pair is named as ***_train.txt and ***_test.txt. 
The first line of these data are always "2 1000".
2 means it is 2-dimensional. 1000 is the number of the data points. The following lines are the data point. 
In training data, each line is two numbers, one for x and one for y.
In test data, each line is two real numbers for x and y, and a intetger to label normal or abnormal: 2 meaning normal, 0 means abnormal.

Within the 38 pairs are seven families of gemetric shapes. 
Shape "Comb" has only one size. 
Each of the other six shapes has three variation of size or parameters, named ***-mid, *-small, ***-big, etc.
1+3*6 = 17 shapes.
Each shape also has it complementery shapes, named ***neg. Thus the total data sets are 17*2=38.