Normal Distribution

I have started attending an 8 month Post Graduate program at Great Learning in Data Science and Business Analytics. While I am hardly able to spend anytime and barely completing videos and sessions, and somehow clearing the quiz - it’s a rewarding experience.

I try and relate the learnings into day to day life, into the business and client problems and sometimes to life at large.

This week’s topic in the course has got me thinking about application of normal distribution. It seems, as per Central Limit Theorem, that if you take a large enough data set - it tends to always follow normal distribution in statistical terms. This, as I think back, goes well with bell curve or normalization curve that organizations apply during appraisal cycle. Good or bad, if we go with this theorem, by and large - the performance of employees in a reasonably large organization, would follow normal distribution, and hence, would fit in a bell curve. Now, due to human biases, difficulties in fair judgement, relative rating, emotions and fear of loss - the actual data doesn’t necessarily follow bell curve, so the organizations tend to apply “force normalization” through rules of the road, roadshows, expectation framework and so on. This, for the most part, works but in a few areas - it doesn’t. Some orgs don’t prefer calling it normalization, give it different terms - but at the end of day, there is some force fitting because data science says (and has proved) that Central Theorem works.

My thoughts may evolve as I get deeper (if i can find time) into this subject. Until then - lets stay ahead of the curve.

(on 18 March 2023) And just for fun - i asked ChatGPT3 to improve the above with better formed sentences… and here goes that version! No doubt it’s better written. It may not bring the “written from heart” emotion but is generally easier to read.

I have recently started an 8-month Post Graduate program in Data Science and Business Analytics at Great Learning. Despite having limited time to complete the videos and sessions, and only just passing the quizzes, the experience has been highly rewarding.

I try to apply what I learn in the course to real-life situations, including business and client problems, and even to life in general. This week's topic in the course, which focuses on the application of normal distribution, has got me thinking about its relevance in statistical terms. According to the Central Limit Theorem, a large data set tends to follow a normal distribution. This concept aligns with the bell curve or normalization curve that organizations apply during appraisal cycles.

In theory, the performance of employees in a reasonably large organization would follow a normal distribution, and hence fit within a bell curve. However, due to human biases, difficulties in fair judgement, relative rating, emotions, and fear of loss, the actual data does not necessarily follow this pattern. Therefore, organizations tend to apply "force normalization" through rules, roadshows, expectation frameworks, and other methods to align the data with the expected pattern. While this method generally works, there are instances where it may not be successful. Some organizations may not even refer to it as normalization, but regardless, there is a need for force fitting because data science has proven the Central Theorem to be accurate.

As I delve deeper into this subject, my thoughts may evolve. For now, it's important to stay ahead of the curve.