The unique property of cluster mapping is observed on a different platform on SPSS software. The intricacy and the beauty of the software, makes it very logical and crisp to understand the concept of the cluster mapping. To put in simple layman’s language, clustering is the processes of putting things with similar features or characteristics under one roof. In other words, all the objects with similar behaviors are grouped under same category to form many mutually exclusive categories.
Though the functionality of cluster mapping may seem complex, it can provide evident statistical results and validated data to derive conclusion from numbers. The correlation and mapping process will collate and process the data using the ordinal values assigned to the variable and eventually represent the understandings of the data.
The K means method of cluster mapping is generally used when the number of cases used for mapping is less than 50 i.e., manageable number of cases that can be studied. Otherwise, hierarchical clustering is used which analyses the data on variables related to the cases. The proximity matrix is a product map which will evaluate the relationships between the factors in terms of fractions.
One of the problems faced by many users is that this analysis always produces clusters, whether there is any underlying structure in the data or not. But K means technique partially solves this problem by iterating from random but strategically choosing starting points.
Thus the common sense judgment of K means helps one to assess the reliability, stability and validity of each analytical solution which can provide evidence of actual cluster structure and between cluster differentiations relative to within cluster similarity.
Perceptual mapping:
This is the technique used to define t he correlation between the elements which are used as variables. The representation output from perceptual mapping will describe the influence of one factor over the other and also determine the change in the performance or behavior of these factors under the influence of other factors. When the variables are plotted against attributes over the vector lines, the nearness of the data points with respect to the vector and their predicted change in behavior with respect to the movement of other factors can be observed. The magnificence of this technique is that the depth of understanding one gets in interpreting the strength of one variable existence over the other.
Perceptual mapping is usually done in 2 steps:
1. Data collection
2. The wide pool of data for perceptual mapping is obtained from rating scales where objects of the map, from populations to products are defined on selected attributes. Therefore choosing right set o attributes to ensure the product classification and identification are appropriate plays a crucial role in obtaining reliable results.
3. Data analysis and presentation
By multiple discriminant analysis as it derives details from between the products and within the products too. The multi discriminant analysis gives way to F ratio which describes the variance between ratings of different objects/ subject to the ratings within objects. This method finds the optimal weighted combination of all the attributes which would produce the highest F ratio between-objects to within-object variations. Unlike cluster analysis, it makes use of attribute ratings rather than similarities and dissimilarities as they are easy and natural to respondents.
These maps provide companies and/ or products relationships on a perceptual space.
These perceptual maps provide insightful observations about the attributes and the objects. These data over years of use, give way to innovative perspectives for a wide range of market research objectives.
Nisha Sullia
2009-11
No comments:
Post a Comment