Prediction Models, Data Analytics RAFA and MEDVEDEV Game

I came across an interesting share on my WhatsApp which triggered some thoughts on data analytics & AI keeping this piece at the focal point. Here are they....?

A) Prediction Models :

I feel we don't need an algorithm to predict what it predicted even we would have said the same thing without a number that Medvedev will win at that point in time in the game but those Rafa fans would always feel otherwise, now that’s the difference between humans and non-human algorithm

Probably what they missed out was as to how many times Nadal has turned around a game from the position that he was in.?

The algorithm should have considered that factor which probably could have reduced the winning probability of Medvedev

B) Machine Learning & AI

To make algorithm more accurate in terms of its predictability it should factor age and correlation between when such a winning result was delivered?

There is no end to what you can factor to build an algorithm to predict certain results. The greater challenge is whether there is a data available or not and how much data is enough, relevant and economical to collect

Other factors could be type of court, type of racket used by the players, the hand of a player (right vs left), crowd in favour and against, time of the day, location of the match, time gap between the matches, etc. There is no end to what data could matter for predictability.?

Each time model will learn new things based on the historical data and show altogether a different result & hopefully its get smarter in terms of predictions it makes.

C) Using refined algorithm (the one that is more learned :-)) to predict the results for the next match and at several points during the match?

This can be done, only challenge is, one has the data for the ongoing match as is needed for algorithm to make accurate predictions. Now in business scenario it means data at several locations, of several machines, of several processes, of several factors (controllable) etc. in short all controllable required to make a prediction

D) Caution in using Data Analytics and AI

We may be surprised one day when prediction model says that there is X% probability of Sun rising from east :-D now in such cases and there will be many such instances where human intelligence and experience should over rule prediction results. After all, we don't want AI to make humans into AI.?

AI is human creation not a vice a versa :-)?

#machinelearning?#dataanalytics?#artificial-intelligence?#innovation?