Math

[EvanLemire]

modz - Takes a pandas df, makes a weighted average of all columns

robust_zscore - Takes one or two pandas dataframes. If a second is provided, it uses this second one for calculating the medians and mads for zscoring (eg. vehicle control)

differential - Use GCToo object metadata to pass either a single dataframe or two dataframes to robust_zscore function based on a given field/value pair

[levlitichev]

Looking good! Are you working on tests? Let me know if I can offer a hand.

[EvanLemire]

@levlitichev Yea I actually already have some tests for these from merino, need to think about how I might change them now that the structure is different / what parts I want to transfer over

[levlitichev]

Made some changes!

• modz was renamed to agg_wt_avg for "aggregate using a weighted average"
• diff_gctoo can do median normalization (i.e. only subtract median, don't divide by MAD) in addition to robust z-scoring
• Various beefing ups of documentation, assertions, and tests

@EvanLemire @oena let me know what you think