The basics of access analysis that you may not know
With the recent spread of remote work,
the importance of web and online marketing measures is increasing.
There is no harm albania cell phone number list in learning the basics of "access analysis" in order to correctly measure the "cluster analysis" introduced in this article.
It is not only important to decide what kind of measures to take, but also to review them after implementation.
Therefore, this document
explains the basics of access analysis, which is essential for verifying the effectiveness of online measures.
Types of cluster analysis
Next, we will explain the main methods of cluster analysis, divided into "hierarchical cluster analysis" and "non-hierarchical cluster analysis."
Hierarchical Cluster Analysis
There are six main methods for performing hierarchical cluster analysis.
Method 1: Ward Method
Method 2: Shortest distance method (nearest neighbor method)
Method 3: Longest Distance Method (Farthest Neighbor Method)
Method 4: Center of gravity method
Method 5: Group Average Method
Method 6: Median method
Method 1: Ward Method
The most commonly used method of cluster analysis in marketing is the Ward method.
In this method, the sum of squares of the distances between multiple data (the sum of squares of the data values and the mean) is the distance between clusters.
Ward's method forms clusters so as to minimize variance, which has the advantage of having very high classification sensitivity and being suitable for data analysis.
However, this method has the disadvantage that it requires a lot of calculations, so care should be taken.
Method 2: Shortest distance method (nearest neighbor method)
The "shortest distance method (nearest neighbor method)" is a method of cluster analysis in which new clusters are formed by using the closest data as the inter-cluster distance.
It has the advantage of requiring less calculations.
However, it should be noted that this method has the disadvantage that if there is data that is extremely distant (outlier), it can easily cause a chain effect, in which similar data is attracted to it and forms a new cluster.
Method 3: Longest Distance Method (Farthest Neighbor Method)
The opposite of the shortest distance method is the longest distance method, which is a method of cluster analysis in which new clusters are formed by using the distance between the furthest data as the cluster distance.
Like the shortest distance method, this method has the advantage of requiring less calculations, but it does have the disadvantage of being vulnerable to outliers and prone to chain effects, although it has a slightly higher classification sensitivity than the longest distance method.
Method 4: Center of gravity method
The method of forming new clusters by using the "center of gravity" between data as the distance between clusters, rather than the distance between data, is called the "center of gravity method."
For example, if you have three pieces of data, the center of gravity between the data is the vertex of a triangle formed by connecting each of the coordinates (A, B, C) and the intersection point between the midpoints of the sides.
It is important to note that with this method the center of gravity will change depending on the number of data contained in the cluster.
Method 5: Group Average Method
The method of forming new clusters by averaging the distances between data from different clusters to determine the distance between clusters is called the "group averaging method."
For example, if cluster "A1" contains data "A and B" and cluster "A2" contains data "C and D," the average of the distances between "A → C and D" and "B → C and D" is the distance between the clusters.
The group averaging method is known as the second most effective method after Ward's method, as it has the advantages of being less susceptible to chain effects and requiring less calculations.