Day 2 Team I Sanjay O S - Few amazing things that simple things like Cross tabs and frequency tests can do.

Started today's class with a simple cross tab analysis, which was built upon gradually to reach to an extent to provide highly detailed insights into the way business functions, fail points and possible areas of improvements with a really simple tool and a good chunk of data to go with it. It all starts with a small business insight of a very superficial data and from where we dig deeper and each such further discovery lead to gaining bigger insight of the problem and narrow the solution with every step. For e.g.: we started analyzing what percentage of customers visit a store and leave without purchasing, which was more or less equal across stores, further we measured the distances traveled by these buyers "not to make the purchase". For two of the four stores the distance traveled by majority customers who didn't make a purchase was less than 5 km whereas in other 2 stores majority who didn't purchase had traveled a distance in range of 10-30 km, this makes very little sense to travel more than 10 km not make a purchase. Further probe suggested that the primary department they were trying to shop in was that of electronics and they not making the purchase can be again attributed to many factors like service level, level of personal contact etc., with this insight our solution and outlook towards the problem is profound and we have only utilized one tool to arrive at it.

It always reminds of the popular quote " Consumers are always right", many times it is the failure of the managers to understand it. Tools like this probably helps us get closer to understand it, "closer".

Next we tried clustering samples,
like every business problem it can best approached by asking one simple question "Why would anyone want to do it?". This leads to the method and extent of clustering to be adopted. and again it is built to make life simple. Came across two methods to do the same agglomerate method (which more of bottom up approach)  and divisive method (which works exactly the opposite way agglomerate method works). Further we looked at how to measure distances between an object and a cluster and between two clusters i.e. different kinds of distance measurement techniques.