Basics

Hello everyone,

As a part of my blog today, I wish to address the commonly faced doubts in the class. We are already past a couple of sessions but some basic doubts seem to prevail. So here we go...

Firstly, regarding the similarity/dissimilarity matrix. The similarity matrix shows the degree of similarity i.e. how closely the two objects under study are related. So when Coke is compared to Coke the Similarity Matrix will show a value of 1 for it. As against this, a Dissimilarity matrix indicates how far apart the two objects under study from each other are. Therefore when Coke is compared to Coke the dissimilarity matrix will indicate a value of 0.

Another confusion was with regards to scale type. The easiest one to recognise is the nominal scale. A nominal scale indicates values in no particular order. The numbers are assigned only to make the data capable of further processing. An ordinal scale is a ranking based scale that aids arrangement. Eg. Our application numbers 61503, 60426. It indicates the student with application number 60426 is previous in the order as compared to the student whose application number is 61503. The third one, scale, is one which is capable of further mathematical solving. For eg. If 5ft 4 inches and 6 feet indicates the height of two people; we can say the difference between their heights is 8 inches.

Today we worked largely on the Permap software. There was a question as to what the objective function is. Also known as error function, it indicates how distant the objects are from a good configuration. When multiple points are plotted on the perceptual map, it may not always be possible to always have the correct distance between them. There is a scope of error which is indicated by Objective Function Value. The software by default minimises this error.

In the Permap the attributes which are being tested for are vectors. The arrows are called attribute vectors which indicates the direction in which the magnitude of that particular vector is increasing. The distance of an object from the arrow indicates how closely it is related to that attribute. Had there been only one attribute which is being tested then all the object would lie on the same vector line. However, since many attributes are tested at once the alignment of objects is adjusted to indicate its closeness correctly to each of the vector attributes.

A lot of us face errors when trying to run the Permap when loading data from a data input file. You could use the following tips to make sure that the errors are resolved.

·         Do not include the headers of the columns when copying from excel too notepad.

·         Correctly number the fields nobjects and nattributes.

·         Remove any spaces that may be there on the objects.

·         Remove any numerals that may be there in the objects.

Eg. More than 1/week must be modified to morethanonceaweek.

I hope these little things will help us in getting more out of the sessions and the software.

Goodbye.