Machine Learning and Product Management

This post in Domino Blog makes a good summary of Pete Skomoroch’s recent Strata London talk.

Probably one of the most important takeaways it’s the fact that ML includes a lot or uncertainties in Product Roadmaps. This kind of uncertainties can be related with model performance, data sources availabilities, tools, processing time, modeling issues and so on. This is highlighted in the following slide:

The point is that more the Product Management tries to enforce some Software Development patterns (that has more deterministic nature) in Machine Learning (that has more probabilistic approach, we will face more and more failed projects in ML.

[slideshare id=145716964&doc=slideshareexecutivebriefingwhymanagingmachinesharderthanyouthink-190515062940]