Well, as always in the field of data and AI, “it depends…😉”
It depends on how informed you are and how informed the algorithm is. Here’s an example to explain:
🏃🏼♀️ As an athlete, I usually follow my Garmin’s data-driven advice. It predicts better than me how fit I feel and what training load I can handle. Pretty awesome, especially if it tells you to rest instead 😊
This time, I overruled it, and here’s why:
📉 November is my decompression month (maybe dragged it on a little longer). The algorithm for training readiness uses the training load in the last month as a base, overestimating its impact.
↪️ I understand the input data are not representative.
🛏️ My watch detected more stress than usual during my sleep, assuming it’s physical stress. If only it knew I’m sleeping in a hostel this week, I’m sure it would understand.
↪️ I have additional information that hasn’t been taken into account.
💟 Based on science and my experience, Heart Rate Variability (HRV) is the best indicator for recovery. This improved last night, so I feel confident it’s safe to train more.
↪️ Applying domain knowledge and experience can improve the prediction.
The same principles apply to data-driven business decisions. They’re often better and less biased than your own. To make good data-driven decisions, the following points are important:
📉 Understand and analyze the underlying data
🔄 Understand the algorithm, its assumptions, and how different factors affect its prediction
🧠 Domain knowledge and expert judgment can improve predictions, especially if you believe assumptions are violated.
So, who was right at the end?
Me, this time. Enjoyed a tough bike ride feeling energized. Next week I’ll listen to my watch again 🍽️🎄
When do you make data-driven decisions? Curious to know! 👇🏼
hashtag#DataDecisions hashtag#DataDriven hashtag#AIinAction hashtag#SportsAnalytics