Using Data Mining to Understand Restaurant Customer Segments

This essay discusses the application of data mining techniques to understand customer segmentation in the restaurant industry. It highlights a scenario where a restaurant chain aims to improve profitability by tailoring marketing strategies and services to specific customer segments. The essay suggests using cluster analysis to group customers based on their perceptions, demographic data, and behavior. It also explains why regression analysis might not be suitable due to potential multi-collinearity issues in large datasets. The goal is to identify key customer segments, understand their characteristics, and determine their contribution to profits, which can be achieved through data mining approaches. Desklib provides a platform for students to access this essay and other study resources.

Running Head: DATA MINING
DATA MINING
Name of Student
Name of University
Author Note

1DATA MINING
Data mining is a specific discipline whereby one makes use of data to recognize
patterns which may be common or even discrepancies to gather insight about various
questions and situations as per the situation (Witten et al., 2016). . Data mining has therefore
become indispensible for businesses and researchers alike for the possibilities that data
mining indicates for their respective purposes.
A typical example of a situation where data mining can be of use is in customer
segmentation in the field of marketing. Consider the example of a management team of a
chain of restaurants is looking to understand the kind of people that visit there branches.
Using insights about tentative market segments they wanted to improve their profits by
designing their marketing strategy and services to suit specific segments which could
optimize their desired business outcomes. Following approaches decribed in Sarstedt &
Mooi (2014), the data can then be imagined as a compilation of customer opinions and
customer data such as income, gender, age, frequency of visit. The data thus includes a series
of customer perception data variables in the form of ordinal ratings scores relating to
questions such as “Comparing prices saves money”, “Eating out is fun”, “ Tasty food trumps
healthy food”, “Vegan diet is a good choice”. The dataset could also include other such
variables which may characterize segments of the market in terms of demography, behaviour,
perception, economic group, etc.
The major question or end goal here is what are the major customer segments of the
restaurant, what are their characteristics and then having identified them it may be discerned
how much each segment may contribute to profits or may hold scope for further business
opportunities. This can be done by identifying common characteristics among the customers
which allows for grouping on the basis of whatever grouping variables with common traits
are used.

2DATA MINING
Following Zhao (2015), clearly then cluster analysis or simply clustering could be a
suitable data mining approach. This can be justified by referring to very nature of the data
and the aim of the research. The data is of ordinal and interval scale which can be rescaled or
used as is to compute distances such as Euclidean distances. Let m variables be used to
discern the clusters then each cluster is denoted by a m-dimensional Euclidean space where
each cluster point is a m-dimensional vectors with elements being the observations from the
cluster grouping variables. Then the distances between the means of the cluster points in
some tentative cluster and that between a point and the mean of a different cluster would help
to determine the arrangement with least variation between groups (or sum of squared
distances) and most between group variation (Sarstedt & Mooi, 2014). The data in this case
can similarly be clustered as per the perceptions of the customers, that is the ratings on the
given statements as mentioned before (Ali & Tuteja, 2014).
Another data mining technique is regression analysis which is also used in many cases
however since data for segmentation is usually very large with large number of attributes and
more often not enough respondents, the data may be subject to multi-collinearity and hence
inflated or uneven variance which makes way for inaccurate predictions making it unsuitable
(Zhao, 2015)

3DATA MINING
References
Ali, S. M., & Tuteja, M. R. (2014). Data Mining Techniques.
Sarstedt, M., & Mooi, E. (2014). Cluster analysis. In A concise guide to market research (pp.
273-324). Springer, Berlin, Heidelberg.
Witten, I. H., Frank, E., Hall, M. A., & Pal, C. J. (2016). Data Mining: Practical machine
learning tools and techniques. Morgan Kaufmann.
Zhao, Y. (2015). Data mining techniques.