more interpretable counterfactual, factual calculation

[m-qin]

The new approach

• Uses the actual events and people affected as the factual; before, we were using our model for the factual
• Uses a threshold exposure (3 miles) for the counterfactual; before, we were generating a matrix of counterfactuals
• Moves all census tracts beneath the threshold to the threshold; before, we were only moving census tracts at least a half-mile away from the threshold

[daniellebraun]

yes, and i think two options to estimate counterfactual events; 1) take factual events at factual exposure apply HR to them using difference between factual and counterfactual exposure 2) take factual events at factual exposure - predicted events at counterfactual (if negative ignore(?))

[m-qin]

Update: deleting the factual_vs_counterfactual R scripts and result files, as they will not be in the manuscript