FSS Aggregated Score Calculation

[malloryprow]

I have a year of NBCRNT lines that I am aggregating. I got

COL_NAME: TOTAL FBS FBS_BCL FBS_BCU FSS FSS_BCL FSS_BCU AFSS AFSS_BCL AFSS_BCU UFSS UFSS_BCL UFSS_BCU F_RATE F_RATE_BCL F_RATE_BCU O_RATE O_RATE_BCL O_RATE_BCU

NBRCNT: 162239416 0.027058 NA NA 0.74922 NA NA 0.99997 NA NA 0.53322 NA NA 0.066944 NA NA 0.066437 NA NA

For the FSS aggregation, are they aggregated in a certain way? They are summed up and then divided by the total? Or what is exactly happening in the aggregation process?

[malloryprow]

2889.tar.gz

[CPKalb]

Hi Mallory:

After some information from @JohnHalleyGotway, the aggregation process proceeds by aggregating Fractions Brier Score (FBS) and the denominator of FSS both using weighted averages. Then, the FSS is recomputed from the FBS and denominator of FSS. It looks like if FSS cannot be aggregated numerically, the value stays the same. Below I'm giving the links to where these calculations are performed in the MET code:

1. NBRCNT is read into Stat-Analysis on line 287 of parse_stat_line.
2. The aggregation is added on line 1450 of aggr_stat_line
3. The actual function that performs the aggregation starts on line 2692 of met_stats

I'm reviewing these calculations to see if everything looks as expected and will post back here once I finish that.

Christina

[malloryprow]

Is this weighted calculation anything I can compute using the values in the NBRCNT line type myself? Are the needed values all in the line type?

[CPKalb]

Hi Mallory:

Yes, all of the needed values are in the NBRCNT line type. I realized I didn't provide a link in the above message, showing where NBRCNT data was read into Stat-Analysis which would have been helpful. I have fixed this, but the data is read into Stat-Analysis on line 287 of parse_stat_line.

The equations used are part of met_stats, with lines 2709 - 2710 calculating FBS, lines 2723 - 2725 for the denominator, and line 2737 for FSS.

The variables in these formulas are as follows:
n_info.sl1l2_info.scount: calculated on line 2697
sl1l2_info.scount: TOTAL from the NBRCNT line
fbs.v : FBS from the NBRCNT line
fss.v : FSS from the NBRCNT line

Anything with a c in front just denotes the current line (versus the line we are aggregating with). Does this all make sense?

When I look at the equations for FSS and FBS in the MET documentation, I think the math used in the code makes sense, but it's also possible I missed something.

Christina

[malloryprow]

Yes! Thank you!