logs/week23.md

In wjones127/thesis: Bike Routes Thesis (Title Case)

Week 23

Missing Data

So assuming MAR a missing response doesn't mean bias in our slope estimates, but it does mean bias in our intercept estimates. This could be especially a problem in estimating the uncertainty we have in the intercepts. We are underestimating our uncertainty of them right now. Maybe.

Datetime is when the ride starts. I just learned that they also have ride end... That would have been useful.

Problems?

I had a thought. what if I randomly assigned riders to rides. Would models perform similarly well if I added a random intercept then? How do I test if those gains were were due to the riders... and not just because I let the model give different sets of observations different values????

wjones127/thesis documentation built on May 4, 2019, 7:34 a.m.