DAY 2- Team F

The Business Analytics class on 4^th September, 2012 started with a recap of what had taken place in the earlier class, and the faculty proceeded to explain us about the different types of scales that are used in the field of Business Analytics. The various types of scales are:

Ordinal Scale- When items are classified according to whether they have more or less of a characteristic, the scale used is referred to as an ordinal scale (definition of ordinal scale). The main characteristic of the ordinal scale is that the categories have a logical or ordered relationship to each other.These types of scale permit the measurement of degrees of difference, but not the specific amount of difference.

Nominal Scale- A discrete classification of data, in which data are neither measured nor ordered but subjects are merely allocated to distinct categories: for example, a record of students' course choices constitutes nominal data which could be correlated with school results.

There was a detailed analysis on the retail store problem that was earlier discussed in the class, and after various permutations and combinations we had to arrive at 2 particular functions/parameters that defined the shopper experience in a retail format. This is where the SPSS software came in handy as it sorted out the different functions and did the mapping accordingly.

The data churned out by the SPSS software had to be carefully analysed for any particular trend to emerge. The different types of analysis that could be put in to use are-

       Cluster Analysis-   Hierarchical (<50)- includes objects like people, brands, etc

2.       K-means (>50) objects are present

Proportionate hazard Analysis- This analysis is used to determine the availability in a particular space, or to see if the particular item fits in a shelf in a retail outlet or not. This again depends on a number of other factors which invariably determines the placement of this particular product in the shelf.

3. Divisional Cluster- A single object in this particular cluster is divided into a number of objects.

After the particular cluster in a group has been identified, it has to be combined in a methodical manner so that it becomes easier for the analyst to take out the meaningful data. One of the most widely used techniques of combining the data clusters is Agglomerative Hierarchical Clustering. It marks every variable as a cluster, and after that two variables having the smallest values are combined into a single cluster. A third cluster is added when the third variable is very less in value, and the number of clusters still remains as 2. This goes on till there are 2 clusters remaining. When you have only 2 clusters remaining, the smallest distance between them is unambiguous. For example, if cluster A has variables 1 and 4, and cluster B has variables 5, 6, and 7, you need a measure of how different or similar the two clusters are.

Clustering criteria:

1.       Where objects are near to each other they can be clustered in to one object called as nearest near criteria.

2.       Distance between the clusters is furthest object.

3.       In centroid clustering we measure distance from center of the objects.

-          Sidharth sivapura

Team F