Carragee Consulting

Do you compete with your partner on who is the “better” driver?

Is the score from your insurance app a valid arbiter in that debate, or does the person with the lower score find ways to justify why the system unfairly penalizes them?

“The app thinks I drove, but I was the passenger when you kept hitting all the red lights!”

“I always have to drive during rush hour and you can drive when there’s fewer cars in the road”

Do drivers Ed teachers show off their Drive Save and Save Scores?
“I’ll teach your kid to drive so well they can game the insurance app!”

Do you ever wish you could play with the data from consumer applications to reverse engineer the ML models that are influential in your life?

Sometimes you can use the public APIs for consumer apps and retrieve your own data, but this is usually of little value.

What you want is your data benchmarked against a valid comparison cohort, which would require access to anonymized data from other users or consent across users to share their data (al la Strava).

Tools like Portable and Peaka are cheap enough to allow individual consumers access to their data from across SaaS applications simply to follow their curiosity without having to write your own APIs.

If I weren’t so brain-addled from Oxy I’d figure how to bring in the relevant part of Kahneman’s book Noise into this stream of consciousness.

As it is I’ll book mark this topic to return to, or you can read it yourself and apply the learnings to your own ML modeling.Activate to view larger image,