Closed bwarren2 closed 8 years ago
Note: the task time limit is much longer than I want because we make 10 s3 requests per filter fn. If this project works, refactoring the pipeline to avoid this bottleneck would be good.
Hopefully the tasks will be faster run from s3, but it is not so bad as things stand.
We have all this per-second time series data for heroes; this turns that into side and side-diff data, rolling things up to the useful levels.
Q: Are the helper functions and list comprehensions too much?