Figure 4B correlation

[trvrb]

In commit f0207d393a30125b24646f92cb7e441516ff39e5 the R correlation coefficient for proportion bad QC in Figure 4B went from 0.22 to 0.01. What's going on here?

This is good reason why informative commit messages are helpful. I can't interpret "updating figure 4". This is also why updating figure specific .ipynb along with updating PNGs is helpful.

[Eabousam]

Fixed issue and reran all data and figures for var of interest

[Eabousam]

Code used to generate the plots are in sections:

• Helper Functions
• Full model data
• Figure 4

[trvrb]

Thanks for looking into this. What's the difference in methodology between the 0.01 result and the 0.22 result? From inspecting b9fec4770c32b1ce9c4e1dfd2d0e224d5d619a65 it looks like the only impactful change would have been swapping mae_seq_full = forecast_mae for mae_seq_full = nowcast_mae. Is this the difference between nowcast vs +30 day forecast error?

[Eabousam]

Yes, 0.01 result comes from filtering the lead == 30 instead of lead == 0.