Day 2 : Team G Malovika Roy BA Work shop

Would consumers make a purchase even if they were unhappy from a store's service?

Does variety satisfaction and gender have any co-relation?

Do people only shop at a particular store because it allows a certain mode of payment?

A lot of such questions got answered today as we analysed the service satisfaction across various stores with cross tabs made against stores and the variety satisfaction and adding a 3rd variable such as mode of payment etc

The results were surprising as a lot of insights were revealed from a business perspective. For example upon selecting a certain store with low consumer satisfaction it was seen that contact with employee was one of the reasons for consumer dis satisfaction. The recommendations given at the time were either to train the employees or adopt best practices from stores which had a higher consumer satisfaction levels after contact with employees.

However, upon further investigation 'A story' was seen to be built around the fact that distance from home and mode of payment were also factors which made for a eason wh people purchased from a particular store.

So the reasons could be many and each team was asked to weave a different story as long as it was substantiated with data.

Now, we come to the part of Cluster Analysis

Cluster Analysis is a statistical technique specifically used in Market research in order to group people , or respondents or cities (all of them can be termed as objects together) based on some common or similar attributes

SPSS has three different procedures that can be used to cluster data: hierarchicalcluster analysis, k-means cluster, and two-step cluster.

If you have a large data file (even 1,000 cases is large for clustering) or a mixture of continuous and categorical variables, you should use the SPSS two-step procedure. If you have a small data set and want to easily examine solutions with increasing numbers of clusters, you may want to use hierarchical clustering. If you know how many clusters you want and you have a moderately sized data set, you can use k-means clustering.

Hierarchical Clustering is of 2 types:-

1. Agglomerative :- Agglomerative hierarchical clustering begins with every case being a cluster unto itself. At successive steps, similar clusters are merged and the algorithm ends with one big cluster.
2. Divisive :-  starts with everybody in one cluster and ends up with everyone in individual clusters. At successive steps individual clusters are split into different clusters

The reason of forming clusters and thus the objective of grouping into clusters should be clear .

To form clusters using a hierarchical cluster analysis, you must select

·         A criterion for determining similarity or distance between cases

·         A criterion for determining which clusters are merged at successive steps

·         The number of clusters you need to represent your data

The DendoGram

For a visual representation of the distance at which clusters are combined, a display called a dendogram is made. In a dendogram visual lines represent the clusters formed in a sequential manner so that formation of clusters can be clearly seen. The observed distances are rescaled however they are done so in the ratio of the actual distances.

The Distance b/w Clusters

• Nearest neighbor (single linkage) : where the nearest objects are combined based on the variables required to map them
• Furthest neighbour : where the furthest objects are groupedtogether
• Centroid Method : where the centroid of the cluster is calculated and distance is mapped from the center of the cluster to another or from cluster to object
• There are other methods for eg Ward’s method, UPGMA method etc