Algorithm 3: SelectK

Input:  Ar{a1,a2…an}, a set of activity

num, number of active classes

Output:k, Optimal k value

1. Val←Ø//

2. Cal←Ø//

3. k←Ø

4. Time←getTime(Ar)//filter out bt and et

5. for each n in range(1, num):

6.    estimator←KMeans(clusters=n)//constructing a cluster

7.    estimator.fit(Time)//fitting the data

8.    Val←getSSE(estimator)//get SSE value

9.    Cal←◡{(n, Val)}

10. end for

11. k←getCur(Cal)//Get k with highest curvature

12. return k