Hello !
Trace: • Media Layers - Wired Network Protocols

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
en:iot-reloaded:clustering_models [2024/09/25 12:30] – agrisniken:iot-reloaded:clustering_models [2024/12/10 21:34] (current) – pczekalski
Line 1: Line 1:
 ====== Clustering Models ====== ====== Clustering Models ======
-{{:en:iot-open:czapka_p.png?50|General audience classification icon}}{{:en:iot-open:czapka_b.png?50|General audience classification icon}}{{:en:iot-open:czapka_e.png?50|General audience classification icon}}\\+
  
 ===== Introduction ===== ===== Introduction =====
Line 6: Line 6:
 Clustering is a methodology that belongs to the class of unsupervised machine learning. It allows for finding regularities in data when the group or class identifier or marker is absent. To do this, the data structure is used as a tool to find the regularities. Because of this powerful feature, clustering is often used as part of data analysis workflow prior to classification or other data analysis steps to find natural regularities or groups that may exist in data.  Clustering is a methodology that belongs to the class of unsupervised machine learning. It allows for finding regularities in data when the group or class identifier or marker is absent. To do this, the data structure is used as a tool to find the regularities. Because of this powerful feature, clustering is often used as part of data analysis workflow prior to classification or other data analysis steps to find natural regularities or groups that may exist in data. 
  
-This provides very insightful information about the data's internal organisation, possible groups, their number and distribution, and other internal regularities that might be used to better understand the data content.  +This provides very insightful information about the data's internal organisation, possible groups, their number and distribution, and other internal regularities that might help us better understand the data content.  
-One might consider grouping customers by income estimate to explain the clustering better. It is very natural to assume some threshold values of 1KEUR per month, 10KEUR per month etc. However:+One might consider grouping customers by income estimate to explain the clustering better. It is natural to assume some threshold values of 1KEUR per month, 10KEUR per month, etc. However:
   * Do the groups reflect a natural distribution of customers by their behaviour?   * Do the groups reflect a natural distribution of customers by their behaviour?
   * For instance, does a customer with 10KEUR behave differently from the one with 11KEUR per month?    * For instance, does a customer with 10KEUR behave differently from the one with 11KEUR per month? 
  
-It is obvious that, most probably, customers' behaviour depends on factors like occupation, age, total household income, and others. While the need for considering other factors is obvious, grouping is not – how exactly different factors interact to decide which group a given customer belongs to. That is where clustering exposes its strength – revealing natural internal structures of the data (customers in the provided example).  +It is evident that, most probably, customers' behaviour depends on factors like occupation, age, total household income, and others. While the need for considering other factors is obvious, grouping is not – how exactly different factors interact to decide which group a given customer belongs to. That is where clustering exposes its strength – revealing natural internal structures of the data (customers in the provided example).  
  
 In this context, a **cluster** refers to a collection of data points aggregated together because of certain similarities ((Understanding K-means Clustering in Machine Learning | by Education Ecosystem (LEDU) | Towards Data Science [[https://towardsdatascience.com/understanding-k-means-clustering-in-machine-learning-6a6e67336aa1]] – Cited 07.08.2024.)). In this context, a **cluster** refers to a collection of data points aggregated together because of certain similarities ((Understanding K-means Clustering in Machine Learning | by Education Ecosystem (LEDU) | Towards Data Science [[https://towardsdatascience.com/understanding-k-means-clustering-in-machine-learning-6a6e67336aa1]] – Cited 07.08.2024.)).
Line 17: Line 17:
   * Cluster **centroid-based**, where the main idea is to find an imaginary centroid point representing the "centre of mass" of the cluster or, in other words, the centroid represents a "typical" member of the cluster that, in most cases, is an imaginary point.   * Cluster **centroid-based**, where the main idea is to find an imaginary centroid point representing the "centre of mass" of the cluster or, in other words, the centroid represents a "typical" member of the cluster that, in most cases, is an imaginary point.
   * Cluster **density-based**, where the density of points around the given one determines the membership of a given point to the cluster. In other words, the main feature of the cluster is its density.    * Cluster **density-based**, where the density of points around the given one determines the membership of a given point to the cluster. In other words, the main feature of the cluster is its density. 
-In both cases, a distance measure estimates the distance among points or objects and the density of points around the given. Therefore, all factors used should generally be numerical, assuming an Euclidian space.  +In both cases, a distance measure estimates the distance among points or objects and the density of points around the given. Therefore, all factors used should be numerical, assuming an Euclidian space. 
  
-===== DBSCAN ===== 
  
-DBSCAN (Density-Based Spatial Clustering of Applications with Noise) employs density measures to mark points in high-density regions and those in low-density regions – the noise. Because of this natural behaviour of the algorithm, it is particularly useful in signal processing and similar application domains.  +==== Data preprocessing before clustering ====
-((Ester, Martin; Kriegel, Hans-Peter; Sander, Jörg; Xu, Xiaowei (1996). Simoudis, Evangelos; Han, Jiawei; Fayyad, Usama M. (eds.). A density-based algorithm for discovering clusters in large spatial databases with noise (PDF). Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96). AAAI Press. pp. 226–231. CiteSeerX 10.1.1.121.9220. ISBN 1-57735-004-9.)). +
  
-One of the essential concepts is the point's p neighbourhood, which is the set of points reachable within the user-defined distance eps (epsilon):+Before starting clustering, several necessary steps have to be performed:
  
-<figure Point's neighbourhood> +  * **Check if the used data is metric:** In clustering, the primary measure is Euclidian distance (in most cases), which requires numeric data. While it is possible to encode some arbitrary data using numerical values, they must maintain the semantics of numbers, i.e. 1 < 2 < 3. Good examples of natural metric data are temperature, exam assessments, and the like—bad examples are gender and colour. 
-{{ {{ :en:iot-reloaded:ClusterEq3.png?300 |  Point's neighbourhood}} +  * **Select the proper scale:** For the same reasons as the distance measure, the values of each dimension should be on the same scale. For instance, customers' monthly incomes in euros and their credit ratios are typically at different scales – the incomes in thousands, while ratios between 0 and 1. If scales are not adjusted, the income dimension will dominate distance estimation among points, deforming the overall clustering results. A universal scale is usually applied to all dimensions to avoid this trap. For instance:  
-<caption> Point's neighbourhood </caption> +     * **Unity interval:** A minimal factor value is subtracted from the given point value and divided by the interval value, giving the result 0 to 1. 
-</figure>+     * **Z-scale:** The factor's average value is subtracted from the original value of the given point and then divided by the factor's standard deviation, which provides results distributed around 0 with a standard deviation of 1. 
  
-, where: 
-  * **p** – the point of interest; 
-  * **N(p)** – neighbourhood of the point p; 
-  * **q** – any other point; 
-  * **distance(p,q)** – Euclidian distance between points q and p; 
-  * **eps** – epsilon – user-defined distance constant; 
-  * **D** – the initial set of points available for the algorithm; 
  
-The algorithm treats different points differently depending on density and neighbouring points distribution around the point – its neighbourhood:+==== Summary about clustering ==== 
 +  * There are many other clustering methods besides the discussed ones; however, all of them, including the discussed ones, require prior knowledge of the problem domain. 
 +  * All clustering methods require setting some parameters that drive the algorithms. In most cases, the value setting might not be intuitive and require interesting fine-tuning.   
 +  * Proper data coding in clustering may provide a significant value even in complex application domains, including medicine, customer behaviour analysis, and finetuning of other data analysis algorithms.   
 +  * In data analysis, clustering is one of the first methods used to acquire the internal structure of the data before applying more informed methods.
  
-  - Core Points: +<WRAP excludefrompdf> 
-    * A point is a core point if it has at least MinPts neighbours within a distance eps, where MinPts and eps are user-defined parameters, i.e. N(p) ≥ MinPts. +To illustrate the mentioned algorithm groups, the following algorithms are discussed in detail: 
-  - Directly Density-Reachable points: +  * [[en:iot-reloaded:k_means|]] - a widely used algorithm that uses distance as the main estimate to group objects; 
-    * A point is directly density-reachable from a core point if it lies within the distance eps of the core point +  * [[en:iot-reloaded:dbscan|]] - a good example of a density-based algorithm widely used in signal processing; 
-  - Border Points: +</WRAP>
-  - Noise points:+
en/iot-reloaded/clustering_models.1727267433.txt.gz · Last modified: 2024/09/25 12:30 by agrisnik
CC Attribution-Share Alike 4.0 International
www.chimeric.de Valid CSS Driven by DokuWiki do yourself a favour and use a real browser - get firefox!! Recent changes RSS feed Valid XHTML 1.0