Scribd Logo
Upload

Welcome to Scribd!

  • 21 views

    DIP Lab 13 DBSCAN Clustering

    Uploaded by

    mahadm.bscs21seecs
    The document describes the DBSCAN clustering algorithm, which can identify clusters of arbitrary shapes in a dataset based on density without specifying the number of clusters. It explains the key parameters and steps of the DBSCAN algorithm, including defining core, border and noise points, and illustrates DBSCAN clustering with example figures. Students are instructed to perform DBSCAN clustering on their own dataset and include code, results and discussion in a lab report.

    Copyright:

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

    DIP Lab 13 DBSCAN Clustering

    Uploaded by

    mahadm.bscs21seecs
    21 views6 pages
    The document describes the DBSCAN clustering algorithm, which can identify clusters of arbitrary shapes in a dataset based on density without specifying the number of clusters. It explains the key parameters and steps of the DBSCAN algorithm, including defining core, border and noise points, and illustrates DBSCAN clustering with example figures. Students are instructed to perform DBSCAN clustering on their own dataset and include code, results and discussion in a lab report.

    Original Description:

    Original Title

    DIP Lab 13 DBSCAN clustering

    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 the DBSCAN clustering algorithm, which can identify clusters of arbitrary shapes in a dataset based on density without specifying the number of clusters. It explains the key parameters and steps of the DBSCAN algorithm, including defining core, border and noise points, and illustrates DBSCAN clustering with example figures. Students are instructed to perform DBSCAN clustering on their own dataset and include code, results and discussion in a lab report.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    21 views6 pages

    DIP Lab 13 DBSCAN Clustering

    Uploaded by

    mahadm.bscs21seecs
    The document describes the DBSCAN clustering algorithm, which can identify clusters of arbitrary shapes in a dataset based on density without specifying the number of clusters. It explains the key parameters and steps of the DBSCAN algorithm, including defining core, border and noise points, and illustrates DBSCAN clustering with example figures. Students are instructed to perform DBSCAN clustering on their own dataset and include code, results and discussion in a lab report.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    of 6

    Department of Electrical Engineering

    Faculty Member: Date:

    Semester:

    Digital Image Processing


    Lab 13: DBSCAN

    Lab Report Quiz/viva


    Name Reg. 10 Marks 5 Marks
    No

    Lab#13: DBSCAN
    Objectives
    This laboratory exercise is focused DBSCAN clustering which is a widely used
    unsupervised learning technique. Clustering is used on unlabeled data to look
    for interesting groups and patterns.

    Lab Instructions
     This lab activity comprises of following parts: Lab Exercises, and Post-Lab
    Viva/Quiz session.
     The lab report shall be uploaded on LMS.
     Only those tasks that are completed during the allocated lab time will be credited
    to the students. Students are however encouraged to practice on their own in spare
    time for enhancing their skills.
    Lab Report Instructions
    All questions should be answered precisely to get maximum credit. Lab report must
    ensure following items:
     Lab objectives
     Python codes
     Results (graphs/tables) duly commented and discussed
     Conclusion
    DBSCAN (Density-Based Spatial Clustering of Applications with Noise)

    DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a


    popular clustering algorithm in machine learning and data mining. It is
    particularly useful for identifying clusters of data points in a dataset based on the
    density of points in the feature space. Unlike k-means or hierarchical clustering,
    DBSCAN doesn't require specifying the number of clusters beforehand and can
    discover clusters of arbitrary shapes.
    Here is the basic algorithm for DBSCAN:
    1. Input:
     Dataset: D={x1,x2,...,xn}, where xi is a data point in the feature space.
     Parameters:
     Epsilon (ε): The maximum distance between two points for one to
    be considered as in the neighborhood of the other.
     MinPts: The minimum number of points required to form a dense
    region.

    Exploring two concepts known as Density Reachability and Density Connectivity


    helps in understanding these parameters.

    Density Reachability, with respect to density, defines a point as reachable from


    another if it is within a specific distance (epsilon) from it.

    Density Connectivity, on the other hand, employs a transitivity-based chaining


    approach to ascertain if points belong to a specific cluster. For instance, points p
    and q may be connected if p->r->s->t->q, where a->b signifies that b is in the
    neighborhood of a.

    2. Algorithm:
     For each data point p in the dataset D:
     If p is not visited:
     Mark p as visited.
     Find all points in the ε-neighborhood of p (including p).
     If the number of points in the neighborhood is less than
    MinPts, mark p as noise.
     Otherwise, create a new cluster and add p to the cluster.
     Expand the cluster by adding all reachable points in the ε-
    neighborhood to the cluster.
    3. Output:
     The algorithm identifies clusters of data points and marks some points
    as noise if they don't belong to any cluster.
    In the algorithm, a point q is considered to be in the ε-neighborhood of p if the
    distance between p and q is less than or equal to ε. The algorithm classifies points
    into three categories:
     Core points: Points with at least MinPts points in their ε-neighborhood.
     Border points: Points with fewer than MinPts points in their ε-neighborhood
    but are reachable from a core point.
     Noise points: Points that are neither core nor border points.
    Figure 1:Credit https://www.theaidream.com/post/dbscan-clustering-algorithm-in-machine-learning

    DBSCAN has advantages such as being robust to outliers and capable of


    discovering clusters of arbitrary shapes. However, it may struggle with datasets
    of varying densities, and choosing appropriate values for ε and MinPts can be
    challenging.
    Figure 2: Credits: https://github.com/NSHipster/DBSCAN

    Lab Task – Your Own Dataset ___________________________________________________


    Download your own CSV dataset from the internet (e.g. Kaggle). Perform
    DBSCAN clustering of your dataset and showcase the plots .
    Lab Task 6 – Take home(optional)
    Download your own CSV dataset from the internet e.g heatmap. Perform
    Hierarchical clustering of your dataset and showcase the plots .

    You might also like