Aggregation Methods for p(activity) vs raw activity

[julie-owen]

@PatWalters I really enjoyed your recent BindingDB presentation and reviewed your example code. I've been using Geo_mean to aggregate raw values (EC50, IC50) where the raw values are log based. But use arithmetic mean to aggregate p(activity) values. I noticed you use Geo mean in your notebook for the p(activity) values. Can you explain why that's a better method than arithmetic mean for p(activity)?

[PatWalters]

@julie-owen you're right, we had already log-transformed the data, so there's no point in using the geometric mean. I've updated the notebook to use just the mean.

[julie-owen]

Thanks for the update here.