So I did end up buying the 52" 3D TV :) The image on the left is of a lonely panda (something I was mirroring from youtube on to my new TV the other day). Truth to be told, I don't watch it as much as I would have liked to.
I had not factored in "free time" when I had done the pareto analysis. One might argue that I had not provided the right amount of input and hence my analysis might have resulted in an inefficient bias reduction.
However, I really doubt that the end result would have changed significantly even if I had in fact provided that information. And here is why:
1. A frequency of the number of times a particular instance happening is not necessarily something that dictates it's perceived importance. For e.g. if you are trying to understand if it is time to sell your old car, and if punctures are a repeated occurrence you probably might want to understand if your tyres are balding or you drive through a construction site everyday ( perhaps you might want to just change the tyres or your route or keep a watchdog).
2. Given it solely relies on past data it cannot account for future anomalies.
Thus, even if I had factored in "free time" and if I had plenty of that, but for some reason ended up being assigned to a new project that demanded a high amount of travel, then I would have landed up in a similar situation.
3. Often times one would try to apply pareto analysis wrongly. For e.g. in ecommerce sites if one ranks items basis their revenue, and tends to disregard items just because their revenue is not as much, and does not evaluate things on a qualitative basis, in the slightly long run this might end up hurting their entire reputation purely basis customer perception.
4. Modelling - it is extremely important that you arrive at a calculated model that ensures accurate scores / weights are getting attached. In the absence of that one might end up running down the wrong path.
More on how to model and assign scores in later posts.
Posts
No comments:
Post a Comment