Ready to dive DEEP into predictive modeling? You’ve come to the right podcast. In this episode, Ben and Michael sit down with Maarit Widmann, a data scientist whose bread and butter is making models more accurate. They discuss how to effectively use confusion matrices and other tools, why you need to avoid THIS misconception to get accurate churn rates, and the BIG question you should be asking if your data seems off.
“If your model is failing, it’s probably been adding up for two weeks already. Instead of monitoring the accuracy, monitor the features. Ask ‘is the data changing?’”
- Maarit Widmann
In This Episode
1) The THREE techniques to help simplify these advanced concepts for beginners in the ML space (and why they need to know this stuff!)
2) Why you need to remember THIS misconception to avoid getting inaccurate churn rates from your models
3) How to effectively use confusion matrices, down sampling, and other popular tools for your models (and the biggest mistakes developers are making today)
4) The REAL question you need to ask yourself if your data seems off (and how this question helps prevent fraud!)
Special Guest: Maarit Widmann.