Description: Colorado%20Tech%20Logo
Colorado Technical University
Applied Managerial Decision – making
MGMT600-1701-CS33-01
Scott, Charlotte
Individual 04 By: Akram Shebani
Colorado Springs, Colorado
Introduction
A lot of businessmen use multivariate techniques for analyzing the responses of the customers by differentiating them into clusters of information or meaningful categories. This is done to study the relationship between the various responses by the consumers and the product or service being studied. Talking about consumer responses, these can be classified into three different categories of dependent, interdependent and independent (Jobson, 2012). Independent responses of the consumer are not affected or swayed by the various variables or conditions, such as the age or gender of the consumer, which is not going to change due to the various characteristics or conditions of the product or service. A dependent response by a consumer is always swayed by an independent response. For instance, a consumer’s preference for a particular video game will be dependent on his age or gender. An interdependent response on the other hand is that in which one particular factor or groups of factors do not characterize the response as a dependent or independent one. For instance, a person’s preference for computers is a multilayered one which is not linked to an isolated factor like age, gender etc. Thus, such a response can be termed as an interdependent response (Gnanadesikan, 2011). The technique using multivariate factors that analyze the data collected from market survey depends upon the relationship between these factors.
Choice of Multivariate Technique: Cluster Analysis
There are three most commonly used techniques for the analysis of interdependent responses, which are factor analysis, cluster analysis, and multidimensional scaling technique. Independent responses can be used as components in every group while making the analysis. In this report, the choice of multivariate technique is cluster analysis. Cluster analysis is the act of segregating a large population into smaller groups of mutually exclusive people based on their common characteristics (Chatfied and Collins, 2013).
Clustering, in simple words, can be defined as the task of categorizing or grouping a certain set of objects that share a similar sense than those objects in other groups. It is the fundamental task when mining for exploratory data. One of the most common techniques when doing statistical analysis of data, it is used in many fields like pattern recognition, information retrieval, machine learning, image analysis, data compression, bioinformatics and computer graphics (Anderberg, 2014).
The technique of clustering has been applied to a wide range of research problems. A summary of the various studies that report the use of the clustering technique have been provided. For instance, when it comes to the field of medicine, clustering the various diseases along with their symptoms and cures can result in useful taxonomies (Murtagh and Heck, 2012). In the field of psychiatry, successful therapy can be achieved with the correct diagnosis from a cluster of symptoms like paranoia, schizophrenia etc. archeology applies the cluster techniques to establish various taxonomies for funeral objects, stone tools etc. In layman’s terms, cluster technique is useful when a mountain of varied information needs to be categorized into meaningful piles of information.
Difference between Cluster Analysis and other Multivariate Techniques
Factor analysis:
When there are a number of interdependent survey responses from consumers that need to be correlated and analyzed, the data collected must be reduced to a group of smaller sets of three or five factors each. As an example of this, when a group of women is being analyzed, their characteristics in reference to their weight, height, interests, hobbies and activities should be studied. If factor analysis is used to study the above characteristics of women, the set of factors would be reduce to two variables of size and lifestyle where size would include a combination of weight and height and lifestyle would use a combination of interests, hobbies and activities (Abdi and Williams, 2010).
Wal-Mart used factor analysis to acknowledge a series of issues they had faced back in 2006 after which they bounced back as the leaders in competitive market nation-wide for their low price product services. Some of the issues faced by Wal-Mart back in 2006 were exporting the jobs and being responsible for destroying small businesses; tax evasion and internal fraud on top level ; being held responsible for class action discrimination suit and litigation; being held responsible for cheap business practices such as cheap health care benefits to employees and low pay scale.
Multidimensional Scaling:
With the help of techniques like multidimensional scaling that records and maps the responses of consumers on a graph as per their similarities, different responses by different consumers can be analyzed and evaluated based on their propinquity to various other responses (Young, 2013). This multivariate technique has been used by Sam’s club to help them determine the neurological preferences and choices of their logo candidates. This helped them select and determine simple steps in order to save their green logo campaign which could identify the various products in their stores that can be termed as environmentally-friendly by basing the decision on the process through which they were created or manufactured, processed and distributed across stores.
In business to business markets, there is little difference found between suppliers and products. To understand the truth behind this one can simply apply the technique of multidimensional scaling. For instance, when a person has five different chemical companies to differentiate between, they can create a questionnaire aimed at the buyers of chemicals made by those companies (Borg et al., 2012). Here each of those five companies would have to be pitted against each other and the respondents would have to answer on scale as to how similar they are on a variety f aspects. How alike their quality of products is? What difference there is in their technical service? How alike their prices are? What is the difference in their quality of product? Once these questions are answered on a scale, a statistical program is used to map these various companies and the factors are tested to understand the similarities and differences between them. These maps are very useful when analyzing the brand to determine if one needs to work on the image of the brand to make it more distinctive in the market.
Real company that used Cluster Analysis:
Cluster analysis was used by SAB miller which is a leading international brewery for beer and also a producer of products for Coca Cola. Cluster analysis helped SAB miller to create classification for all their liquids in order to create easy flow of communication for both technical staff as well as marketing personnel. It also helped the consumers to understand the exact differences between the different beers produced in the brewery which in turn helped the company to understand and determine the preferences of the consumers (Mooi and Sarstedt, 2010). Cluster analysis also helped the company to identify and acknowledge the various opportunities for their brand in the market along with being able to determine the acceptability of their particular brand in the international and local markets.
Companies ideally use clustering techniques to segment the products, customers and store. The above report talks in detail about customer segmentation by using the cluster technique. In the same manner products of a company can also be categorized into different clusters or groups based on their brand, flavor, use, size etc. even stores with similar characteristics based on customer sales, similar sales, size etc can be grouped together (Duran and Odell, 2013). The clustering technique is also very good for detecting any kind of anomaly like identification of theft or fraud transactions. This can be done by creating a cluster or group of good transactions. Whenever a transaction falls out of that group for whatever reason it creates an anomaly and thus can be termed as a suspect which is then detected to see where the transaction went wrong. This can help identify certain fraud transactions. This clustering approach is highly useful in the field of medicine where one can detect any abnormal cells in tissue samples (Anderberg, 2014). It can also be used to detect various calling patterns in telecommunication companies to identify fraudulent behavior. The clustering technique is usually used to break down a complex set of varied information into amenable piles of data that are more acquiescent to other techniques. For instance, logistic regression outcomes can be better performed in small clusters that follow different distributions and behave differently.
Cluster Analysis used for Market Segmentation in my Organization
There are various ways to divide and segment a market but one of the most effective and precise ways to segregate a market in a statistically valid manner would be to use the cluster analysis technique. The tool of cluster analysis is used in various disciplines and not just the marketing field (Wedel and Kamakura, 2012). Basically, any place where there is lots of data to be segregated and condensed into smaller pools of information or clusters are called clustering of data, what marketing segments mean in the field of marketing. When one tries to group the consumer data based on their needs, behaviors, attitudes etc into related sets, it is called the market segmentation process. To create this segmentation the cluster technique is used to review and create such market segments.
In my organization i.e. retail store for apparels and footwear, market segmentation is very important. It is important for me to understand who my customers are and where their loyalty lies. While I used the basic demographic, behavioral and psychographic segmentation before, now I believe that with cluster analysis I will be able to segment the market on the basis of customer satisfaction and loyalty. This will help me focus on more specific customers and accordingly help in make proper marketing strategies and gain more customers (Tuma et al., 2011). The following is how I would use cluster analysis for market segmentation in my company.
Customer base is segregated in the market based on two major variables, which are customer satisfaction and loyalty metric (Cleveland et al., 2011). Assuming we have a customer satisfaction (CSAT) score ranging from 1 to 9 where 1 means very dissatisfied to 9 which means highly satisfied. Similar scores for customer loyalty are 1, which means high switcher-low loyalty, and 9 which means low switcher-high loyalty.
The above graph shows the customer database as a scatter-plot graph where the red squares represent the various scores of the individual customers and the red circle in the middle represents the average score of all the customers for CSAT (Muller and Hamm, 2014). When you look closely at the above diagram one can notice an innate pattern that identifies three inherent market segments as shown in the chart below.
The above chart clearly shows that there are three segments of clusters for consumers as suggested by the data collected. The black circle on the top right corner represents the set of customers who are highly loyal towards the company and its products and services. These customers are loyal with a high level of customer satisfaction. The blue circle in the graph on the bottom left corner appears to represent the customers who are least loyal towards the company’s products and services (Pritchard and Howard, 2015). These customers are less loyal and have a lower level of customer satisfaction. This sows a relationship between the customers and managers, which are fairly obvious and expected where marketing programs, are built around this particular customer satisfaction correlation. When you observe the red circle on the top left corner of the graph you will find that this area represent customers that are highly dissatisfied yet they are loyal towards the products and services of the company (Dolnicar et al., 2014). This is a highly interesting find and mostly unexpected. This is why it is very important to look at the various approaches of the market segments.
Conclusion
To conclude, there are many multivariate techniques however, clustering is a great way to study the various patterns and structures in the database so as to understand their scope of application in business analytics. There are many techniques and different ways to perform the clustering of data. As an analyst one should have proper knowledge and expertise as to how to perform clustering along with having an insight into the various clustering algorithms. The analyst should be able to apply the technique that is most relevant as per the needs of the business. Cluster analysis is best when used for segmenting the market and hence I selected this tool for my organization.
References
Abdi, H. and Williams, L.J., 2010. Principal component analysis. Wiley interdisciplinary reviews: computational statistics, 2(4), pp.433-459.
Anderberg, M.R., 2014. Cluster analysis for applications: probability and mathematical statistics: a series of monographs and textbooks (Vol. 19). Academic press.
Borg, I., Groenen, P.J. and Mair, P., 2012. Applied multidimensional scaling. Springer Science & Business Media.
Chatfied, C. and Collins, A.J., 2013. Introduction to multivariate analysis. Springer.
Duran, B.S. and Odell, P.L., 2013. Cluster analysis: a survey (Vol. 100). Springer Science & Business Media.
Cleveland, M., Papadopoulos, N. and Laroche, M., 2011. Identity, demographics, and consumer behaviors: International market segmentation across product categories. International Marketing Review, 28(3), pp.244-266.
Dolnicar, S., Grün, B., Leisch, F. and Schmidt, K., 2014. Required sample sizes for data-driven market segmentation analyses in tourism. Journal of Travel Research, 53(3), pp.296-306.
Gnanadesikan, R., 2011. Methods for statistical data analysis of multivariate observations (Vol. 321). John Wiley & Sons.
Jobson, J.D., 2012. Applied multivariate data analysis: volume II: Categorical and Multivariate Methods. Springer Science & Business Media.
Mooi, E. and Sarstedt, M., 2010. Cluster analysis. In A concise guide to market research (pp. 237-284). Springer Berlin Heidelberg.
Müller, H. and Hamm, U., 2014. Stability of market segmentation with cluster analysis–A methodological approach. Food Quality and Preference, 34, pp.70-78.
Murtagh, F. and Heck, A., 2012. Multivariate data analysis (Vol. 131). Springer Science & Business Media.
Pritchard, M.P. and Howard, D.R., 2015. Measuring loyalty in travel services: A multi-dimensional approach. In Proceedings of the 1993 World Marketing Congress (pp. 120-124). Springer, Cham.
Tuma, M.N., Decker, R. and Scholz, S., 2011. A survey of the challenges and pitfalls of cluster analysis application in market segmentation. International Journal of Market Research, 53(3).
Wedel, M. and Kamakura, W.A., 2012. Market segmentation: Conceptual and methodological foundations (Vol. 8). Springer Science & Business Media.
Young, F.W., 2013. Multidimensional scaling: History, theory, and applications. Psychology Press.