Skip to main content

Table 11 Refinement process

From: On K-means clustering-based approach for DDBSs design

Cluster

Cluster members (overlapping clusters)

P#

Partitioning schemes

Missing partition (added)

Schemes combinations Non-overlapping PS

PS state

Optimality measure (OM)

Decision making

CQ235

Q2(A3,A5)

Q3(A2,A4,A5)

Q5(A1,A2,A5)

P1

(A3,A5) (A2,A4) (A1)

A6

(A3,A5) (A2,A4)

(A1,A6)

Kept

37.5%

Exclude

P2

(A2,A4,A5) (A1,A3)

(A2,A4,A5) (A1,A3,A6)

Kept

62.5%

Include

P3

(A1,A2,A5) (A3,A4)

(A1,A2,A5) (A3,A4,A6)

Kept

62.5%

Include

P4

(A1,A2,A5,A6) (A3,A4)

Kept

75%

Include

CQ14678

Q1(A1,A2,A3,A6)

Q4(A1,A3,A6)

Q6(A3,A4,A6)

Q7(A2,A6)

Q8(A1,A3,A5,A6)

P5

(A1,A2,A5,A6) (A3,A4)

None

Replica of P4

Deleted

–

–

P6

(A1,A3,A6) (A2,A4,A5)

Replica of P2

Deleted

–

–

P7

(A3,A4,A6) (A1,A2,A5)

Replica of P3

Removed

–

–

P8

(A2,A6) (A1,A3,A5) (A4)

(A2,A6) (A1,A3,A5) (A4)

Kept

37.5%

Exclude

P9

(A1,A3,A5,A6) (A2,A4)

(A1,A3,A5,A6) (A2,A4)

Kept

75%

Include