In this example, we would rule out the two market segment structure, as its SSE is too high relative to the other potential approaches – it is almost double the error of the five segment approach for instance. Therefore, we need to accept a degree of error – (where “error” simply means that the consumers in each market segment are not completely identical, they will vary to some extent in terms of some of their behaviors and psychographic factors). For example, if we had 100 consumers allocated to 100 different market segments, then the SSE would be zero – but, of course, it becomes impractical for a firm to pursue 100 market segments and, as well, each market segment would be too small to be viable. (This SSE chart is automatically produced by the template – you will find it on a separate worksheet – please see the tabs at the bottom of the spreadsheet.)Īs you can see, as more market segments are created the SSE improves. To further clarify, let’s have a look at these sum of squared error (SSE) outputs on a graph, as shown here. To make sense of what to look for, let’s consider the following sum of squared error outputs: Using SSE: An example (for SSE by Number of Segments)
Please note that you don’t need to select the segmentation approach with the lowest SSE, but it should be one of the lower ones. A high SSE suggests that the consumers in the same market segment have a reasonable degree of differences between them and may not be a true (or usable) market segment. The lower the SSE, then the more similar are the consumers in that market segment. So we need to look for a segmentation approach that has a lower SSE. If the consumers matched the segment scores exactly, the the sum of squared error (SSE) would be zero = no error = a perfect match.īut with real world data, this is very unlikely to happen. It does this by performing repeated calculations (iterations) designed to bring the groups (segments) in tighter/closer. What is sum of squared error (SSE)?Ĭluster analysis is a statistical technique designed to find the “best fit” of consumers (or respondents) to a particular market segment (cluster). Please check this diagram for where to locate the SSE scores on the Excel template – plus there a separate worksheet in the template that has two SSE charts for quick reference. Sum of squared error, or SSE as it is commonly referred to, is a helpful metric to guide the choice of the best number of segments to use in your end segmentation.Īs the free Excel cluster analysis template automatically provides the output for two, three, four and five market segments – you need to choose the most appropriate number of segments for your marketing situation – and SSE will help you do that to some extent.
Sum of Squared Error (SSE) in Cluster Analysis
0 kommentar(er)
