A Multiscale Clustering Approach for Non-IID Nominal Data

Algorithm 1 Benchmark-scale clustering algorithm (BSCA).

	Input: raw dataset
Output: clustering result of benchmark-scale dataset
1: Data preprocessing and constructing multiscale dataset MD
2: Choose benchmark-scale dataset MDi[j] from MDi = {{},{},…, {}}
3: foreach MDi[j]:
4:  compute matri_W =  according to (4)
5:  AdjD = getAdjD(matri_W)
6:  LnM = AdjD-matri_W
7:  DnM = AdjD^-1/2
8:  LM = DnM ╳ LnM ╳ DnM
9:  value, vector = getEigenvector (LM, N)
10:  clusters = KMeans(N)
11:  st = clusters.fit(vector)
12:  Cl = st.labels_
13:  Cc = st. cluster_centers_
14:  Rcenter[i]j = Cc||Cl
15: endfor
16: return Rcenter[]