Email
sinha.rk.dr@gmail.com
Do you believe in the weather forecasts?
I have asked this question to many people over the past few years and met with a similar reaction. People laughed at the weather forecasts, joking that we must carry raincoats if it’s predicted to be sunny! The reason for poor accuracy was quite simple. There were so many variables which could cause a change in the weather and it was difficult to predict them accurately even in the short term.
However, the true reason was actually quite different!
We got weather forecasts at a city level. Certain large cities had a couple of these forecasts. For example, Mumbai used to have forecasts for Colaba and Santacruz separately. These were aggregate forecasts. Colaba forecast was meant for the entire south Mumbai and Santacruz one was applicable to all the suburbs. However, people in Powai, Ghatkopar, Andheri, Bandra and Borivli experienced different patterns and couldn’t relate to the Santacruz forecast.
Things have changed now.
We can now select ‘Your location’ in the weather app and get a localised forecast, which is quite accurate in the near term. If it indicates ‘light rain’ over the next few hours, we must carry raincoats.
What has really caused the accuracy to improve significantly?
The AI based prediction models are now based on local variables and auto tuned separately for each locality with a self-learning mechanism. As a result, the more granular we go, the better is the forecast accuracy.
Are there learnings here for Supply Chain planning?