Out of Sample ->Out of Mind

Hi everyone, I hope everyone is fine. Today I am going to discuss the classic example of "Out of Sample Data" where our forecasting models can go wrong badly.

Let's begin our Conversation.

George is a highly educated and non-drinker living with his family. He has his car and drove around 2000 successful trips with only two minor accidents.

One night he was frustrated and drunk heavily because his close friend was leaving the company. He has a call with the client the next day early morning, which he can only take if he reaches home fast and takes sufficient rest.

But George knew he was heavily drunk he was about to postpone the meeting and book a taxi to reach home. But somewhere inside his mind, a question arose " I am driving for 10 years with 2000 successful trips never met with an accident Why can't I drive fast myself in my car and attend tomorrow's meeting? "

Let's discuss the situation of George in detail now.

Case 1:

George knew the risk, postpones the meeting and takes a Taxi to drive home.

Happy Ending!

Case 2:

George can brag about his experience and record of 2000 successful trips and convince himself to drive his car to reach home fastly So that he can attend tomorrow's meeting.

George started driving his car but in the midway, he met with a horrible accident.

Horrible ending!

What went wrong?

George was never drunk during his 2000 successful trips. So his past data and experience cannot be used to forecast the future outcome of the current situation. This was an Out of Sample situation.

Plot Twist:

So what do you think George has gone with Case 1 or Case 2. George is a data scientist who was completely aware of this out of sample data, Postponed the meeting, and booked a taxi.

Be like George and be aware of "Out of Sample Data".

References :

The Signal and the Noise: Why So Many Predictions Fail-but Some Don't