DAY 5 - TEAM G

Perception – A strong weapon that could make or break several brands!

After having studied the SPSS software, we have now moved on to newer domains. PERMAP (Perceptual Mapping) is the new software that Team NMORE has introduced to us today. A powerful freeware, Permap could replace the MDS functionality of SPSS. [MDS-Multi Dimensional Scaling]

The 2 possibilities are:

1.       Overall similarity: It is performed for each pair of objects from the data.

a.       Advantages: extremely flexible and yet powerful; ability to find out what we cannot directly tell (unconscious/back of the mind/hidden attributes)

b.      Disadvantages: since the attributes are not specified, we have to find them out by looking at the diagrams. Having thorough knowledge about the object is extremely important here.

2.       Attribute based: It is based on ratings and rankings

a.       Advantages: simpler in comparison to the previous method, especially for the respondent

b.      Disadvantages: it is possible to miss out on certain important attributes

Preliminary syntax for PERMAP:

a.       title=TITLENAME

b.      nobjects=NUMBER.OF.OBJECTS

c.       similaritylist OR Dissimilaritylist

Example for usage of PERMAP:

Consider a case where the distance matrix between cities is being considered. The extracted distance matrix is fed into a notepad file with the initial syntax. This file is then loaded into PERMAP.

 

File>Load data from a data input file (F2)

After loading the notepad file, click on START. This would give an initial clustering outlay in the output portrait. By clicking once outside the circle, we can see the names of each object.

Note: The “Objective Function Value” is the error in the method. This could arise due to various reasons and could be reduced in several ways.

Check the boxes for “Auto Repeat” and “Auto Stop” and click on Start again. Once the Objective Function Value has settled for over 10 iterations (or 5 seconds), click Stop. This would be the minimum value of error attainable under the current scenario.

Right click inside the window to gain access to a new set of controls. Here, we can use Mirror Click, Rotate Drag, and Zoom Drag etc. to adjust the primary aesthetics of the display. By right clicking again, we go back to the former set of control buttons.

To the left of the circular display, we can see the On-line controls. Badness refers to the tightness of the clustering process. For most business applications, our functionality is to be restricted to the Dimensions alone. By changing the dimensions from 2 to 3 or higher, we should observe any changes in the error value (Object Function Value).

The box to the right of the circular display named Parked Objects allows us to experiment with the clustering process. We can check the impact of removal of one or more objects from the data by dragging and dropping the object into this space.

View>show coordinate axis: This draws the axis lines inside the circular space. These axes could be allocated to specific parameters. For example, if the data set were about specific products, the X and Y axes could be assigned to price and quality factors.

Author:

Anand Chandran
Team G