Scribd Logo
Upload

Welcome to Scribd!

  • 208 views

    Objective: For One Dimensional Data Set (7,10,20,28,35), Perform Hierarchical Clustering

    Uploaded by

    FARHAN KAMAL
    The document describes performing hierarchical clustering on a 1D data set using agglomerative hierarchical clustering (AGNES) and plotting the dendrogram. It performs both single and complete linkage clustering. For single linkage, it forms two clusters - (7,10) and (20,28,35). For complete linkage, it forms two clusters - (7,10,20) and (28,35). It concludes that hierarchical clustering is useful when a hierarchy is required but has high computational and storage costs.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    Objective: For One Dimensional Data Set (7,10,20,28,35), Perform Hierarchical Clustering

    Uploaded by

    FARHAN KAMAL
    208 views13 pages
    The document describes performing hierarchical clustering on a 1D data set using agglomerative hierarchical clustering (AGNES) and plotting the dendrogram. It performs both single and complete linkage clustering. For single linkage, it forms two clusters - (7,10) and (20,28,35). For complete linkage, it forms two clusters - (7,10,20) and (28,35). It concludes that hierarchical clustering is useful when a hierarchy is required but has high computational and storage costs.

    Original Title

    AGNESClustering

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document describes performing hierarchical clustering on a 1D data set using agglomerative hierarchical clustering (AGNES) and plotting the dendrogram. It performs both single and complete linkage clustering. For single linkage, it forms two clusters - (7,10) and (20,28,35). For complete linkage, it forms two clusters - (7,10,20) and (28,35). It concludes that hierarchical clustering is useful when a hierarchy is required but has high computational and storage costs.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    208 views13 pages

    Objective: For One Dimensional Data Set (7,10,20,28,35), Perform Hierarchical Clustering

    Uploaded by

    FARHAN KAMAL
    The document describes performing hierarchical clustering on a 1D data set using agglomerative hierarchical clustering (AGNES) and plotting the dendrogram. It performs both single and complete linkage clustering. For single linkage, it forms two clusters - (7,10) and (20,28,35). For complete linkage, it forms two clusters - (7,10,20) and (28,35). It concludes that hierarchical clustering is useful when a hierarchy is required but has high computational and storage costs.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    of 13

    Objective 

    : For one dimensional data set {7,10,20,28,35}, perform hierarchical clustering


    (AGNES) and plot the dendrogram to visualize it.

    Solution : First, visualize the data.

    By observing the plot above, we can intuitively conclude that:

    1. The first two points (7 and 10) are close to each other and should be in the same cluster
    2. Also, the last two points (28 and 35) are close to each other and should be in the same
    cluster
    3. Cluster of the center point (20) is not easy to conclude

    Let’s solve the problem using both types of agglomerative hierarchical clustering:

    1. Single Linkage: In single link hierarchical clustering, we merge two clusters in each
    step, whose two closest members have the smallest distance.
    Using single linkage, two clusters are formed:

    Cluster 1 : (7,10)

    Cluster 2 : (20,28,35)


    2. Complete Linkage: In complete link hierarchical clustering, we merge those members of the
    clusters, which provide the smallest maximum pairwise distance in each step.

    Using complete linkage, two clusters are formed :

    Cluster 1 : (7,10,20)

    Cluster 2 : (28,35)


    Conclusion: Hierarchical clustering is mostly used when the application requires a hierarchy, e.g
    creation of a taxonomy. However, they are expensive in terms of their computational and storage
    requirements.

    Second Example (Single Link):

    Consider the following points:

    X Y
    P1 0.40 0.53
    P2 0.22 0.38
    P3 0.35 0.32
    P4 0.26 0.19
    P5 0.08 0.41
    P6 0.45 0.30

    Solution: let’s visualize the data.

    Iteration 1:
    The minimum distance is between P3 and P6 (0.11), Therefore, we form first cluster containing two
    points (P3, P6). Then

    Iteration 2:

    Step 3: How distances are re-calculated


    The minimum distance is between P2 and P5 (0.14), so we form second cluster containing two points (P2,
    P5). Then

    Iteration 3:

    Distance {(P2, P5), (P3, P6)}

    = MIN {dist {(P2, P5), (P3, P6)}}

    = MIN { dist(P2, P3), dist(P2, P6), dist(P5, P3), dist(P5, P6)}

    = MIN { 0.15, 0.25, 0.28, 0.39} = 0.15


    The minimum distance is between (P3,P6) and (P2,P5) = (0.15), so we form third cluster containing four
    points (P2, P3, P5, P6). Then

    Iteration 4:

    Distance {(P2, P5, P3, P6), (P1)}

    = MIN {dist {(P2, P5, P3, P6), (P1)}}

    = MIN { dist(P2, P1), dist(P5, P1), dist(P3, P1), dist(P6, P1)}

    = MIN { 0.23, 0.34, 0.22, 0.23} = 0.22


    The minimum distance is between (P4) and (P2,P5,P3,P6) = (0.15), so we form fourth cluster containing
    five points (P2, P3, P4, P5, P6). Then finally we have total two clusters:

    Cluster1 = {P1}

    Cluster2 = {P2, P3, P4, P5, P6}

    ---------------------------------------------------------------------------------------------------------------------
    Second Example (Complete Link):

    Consider the following points:

    X Y
    P1 0.40 0.53
    P2 0.22 0.38
    P3 0.35 0.32
    P4 0.26 0.19
    P5 0.08 0.41
    P6 0.45 0.30

    Solution:
    Iteration 1:

    The minimum distance is between P3 and P6 (0.11), Therefore, we form first cluster containing two
    points (P3, P6). Then

    Iteration 2:

    How distances are re-calculated

    To compute the distances between clusters


    Distance ((P3, P6), P1) = MAX (dist (P3, P6), P1)
    = MAX (dist (P3, P1), dist (P6, P1))
    = MAX (0.22, 0.23)
    = 0.23
    The minimum distance is between P2 and P5 (0.14), Therefore, we form second cluster containing two
    points (P2, P5). Then

    Iteration 3:

    To compute the distances between clusters


    Distance ((P2, P5), (P3, P6)) = MAX (dist (P2, P5), (P3, P6))
    = MAX (0.15, 0.25, 0.28, 0.39)
    = 0.39

    P1 P2, P5 P3, P6 P4
    P1 0
    P2, P5 0
    P3, P6 0.23 0.39 0
    P4 0.37 0.22 0
    Third Example (Average Link):

    Consider the following points:

    X Y
    P1 0.40 0.53
    P2 0.22 0.38
    P3 0.35 0.32
    P4 0.26 0.19
    P5 0.08 0.41
    P6 0.45 0.30

    Solution:
    Iteration/Step 1:

    The minimum distance is between P3 and P6 (0.11), Therefore, we form first cluster containing two
    points (P3, P6). Then

    Iteration/Step 2:

    How distances are computed

    To compute the distances between clusters


    Distance ((P3, P6), P1) = MEAN (dist (P3, P6), P1)
    = MEAN (dist (P3, P1), dist (P6, P1))
    = MEAN (0.22, 0.23)
    = (0.22 + 0.23) / 2
    = 0.225
    Distance ((P3, P6), P2) = MEAN (dist (P3, P6), P2)
    = MEAN (dist (P3, P2), dist (P6, P2))
    = MEAN (0.15, 0.25)
    = (0.15 + 0.25) / 2
    = 0.20

    Distance ((P3, P6), P4) = MEAN (dist (P3, P6), P4)


    = MEAN (dist (P3, P4), dist (P6, P4))
    = MEAN (0.15, 0.22)
    = (0.15 + 0.22) / 2
    = 0.185

    Distance ((P3, P6), P5) = MEAN (dist (P3, P6), P5)


    = MEAN (dist (P3, P5), dist (P6, P5))
    = MEAN (0.28, 0.39)
    = (0.28 + 0.39) / 2
    = 0.335

    P1 P2 P3, P6 P4 P5
    P1 0
    P2 0.23 0
    P3, P6 0.225 0.20 0
    P4 0.37 0.20 0.185 0
    P5 0.34 0.14 0.335 0.29 0

    The minimum distance is between P2 and P5 (0.14), Therefore, we form second cluster containing two
    points (P2, P5). Then

    Iteration/Step 3:

    Distance ((P3, P6), (P2, P5)) = MEAN (dist (P3, P6), (P2, P5))
    = MEAN (dist (P3, P2), dist (P3, P5), dist (P6, P2), dist (P6, P5))
    = MEAN (0.15, 0.28, 0.25, 0.39)
    = (0.15 + 0.28 + 0.25 + 0.39) / 4 = 0.267

    P1 P2, P5 P3, P6 P4
    P1 0
    P2, P5 0
    P3, P6 0.225 0.267 0
    P4 0.37 0.185 0

    You might also like