Day 2 Team C (Rohit)

Day 2 ...The Exciting Voyage Of Business Analytics Subject Continues.....

04/09/2012, Morning 10.30am to 1.15pm

What we learnt?

1. Example taken was Retail from SPSS files. (File name: Retail.sav)

2. Case Selection & its application.

3. Use of “IF” and “&”.

4. Use of Frequencies.

5. Cluster Analysis-its types.

Case Selection & its Application: In order to select case, we identify the two quantities affecting each other. This is purely based on Intuition & Experience.

For e.g.: we took stores in row and service satisfaction in column and tried to establish the relationship between the two. Usage of Chi-Square test was done.

Also, Comparison between stores & contact employees was done. This to find out does contact with employees affect the stores.

Similarly it was done for Gender & primary department (clothing, shoes, appliances etc)

Use of Chi Square test & Crosstab was done to arrive at output systematically.

Each Store feedback was studied to see which has more positive feedback and which has more negative feedback.

Use of “If” and”&”: “If” is used as condition ,once it is satisfied then we are moved to valid cases for that condition, so by this we eliminate irrelevant data & get valid data useful for analysis.

Use of Frequencies: It is used to obtain counts on a single variable's values.

Cluster Analysis: It is a class of technique used to classify cases into group that are relatively homogenous within themselves & heterogeneous between each other on the basis of defined set of variable group are called clusters.

Why & where we use Cluster Analysis?

It used in Market Research for following purposes:

·         Market Segmentation

E.g.: Cluster of consumers according to their attribute preferences.

·         Understanding Buyers Behaviour: Consumer with similar behaviours/characteristics is clustered.

·         Identifying New Product Opportunities: Clusters of similar brands & products can help identify competitors/market opportunities.

·         Reduce Data

E.g.: Preference Mapping

Clustering Procedures:

·         Hierarchical Procedures

Ø  Agglomerative: Start from ‘n’ clusters to get to 1 cluster.

Ø  Devisive: Start from 1 cluster to get to ‘n’ cluster.

·         Non Hierarchical Procedures

Ø  K-.means clustering:  (Note: This is yet to be taught)

·         Hierarchical used for less than 50 objects, k-means used for greater than 50 objects.

·         Dendogram: It is graphical representation of clusters & how they are similar in terms of attributes, characteristics and preferences. It creates one cluster at the combination of different clusters.

·         Distance Measurement: It consists of Interval, counts & binary.

·     

    Clustering Criteria 

Ø  Nearest & Furthest Neighbour depending upon nearness & farness of the objects.

Ø  Centroid Clustering based on distance from the centre of the objects.

Hierarchical	Non Hierarchical
No decision about the number of clusters.        Problems where data contains high        level of error.          Can be very slow     Initial decision are more influential(one step only)	Faster & more reliable. Need to specify the number of clusters(arbitrary) Need to set the initial (arbitrary)

Recommended Approach for arriving at precise & more close outputs:

·         First perform Hierarchical Procedures to define number of clusters.

·         Then use K-means procedure to actually form the clusters.

 By Rohit Thorat 

Team C

References:

http://www-users.cs.umn.edu/~kumar/dmbook/ch8.pdf

https://statistics.laerd.com/spss-tutorials/testing-for-normality-using-spss-statistics.php