Vecna
a8a0983f9e
Use statrs functions
2024-05-27 18:55:37 -04:00
Vecna
11bedfb74a
Only use CDF tests if today's numbers are worse than average
2024-05-27 17:47:02 -04:00
Vecna
659b8fa16c
Refactor analysis for efficiency
2024-05-22 16:04:52 -04:00
Vecna
d9aa616d77
Use estimated multivariate CDF when we have positive reports
2024-05-21 19:30:09 -04:00
Vecna
3512adc425
Stage 2: Model as two distributions, handle 0 standard deviation
2024-05-21 17:16:26 -04:00
Vecna
97d4622cd4
Use CDF, not PDF with artificial 'blocked' data
...
TODO: Figure out proper multivariate CDF
2024-05-20 20:54:28 -04:00
Vecna
22163cc030
Only re-evaluate new data, analyze based on user-defined interval
2024-05-20 20:38:06 -04:00
Vecna
b50f40fe8a
Add noise when necessary to build distribution
2024-05-04 14:26:18 -04:00
Vecna
c38b182cb7
Remove unused import
2024-04-27 14:32:26 -04:00
Vecna
e2b1f9aa7d
Start stage 3 based on actual observation of a positive report
...
Also track how much historical data we have on a per-country basis
2024-04-27 13:20:05 -04:00
Vecna
10ff088b5a
Resolve some TODOs
2024-04-26 15:25:56 -04:00
Vecna
31b27a291a
Model both accessible bridge and blocked bridge in analysis
2024-04-26 15:09:30 -04:00
Vecna
b2eb244757
cargo-fmt
2024-04-26 13:11:32 -04:00
Vecna
273aaab38d
Don't scale bridge IPs in analysis
2024-04-26 12:58:03 -04:00
Vecna
e889cba878
Remove info print statements
2024-04-19 01:34:24 -04:00
Vecna
851686cf94
Mark unused variable bridge_ips in stage one analysis
2024-04-18 22:42:27 -04:00
Vecna
36395181d3
Improve analysis
...
We seem to get better results when we scale bridge IPs down to multiples of 1 instead of 8 and only mark bridges as blocked if they differ in the 'bad' direction by at least one standard deviation from the mean.
2024-04-18 22:27:57 -04:00
Vecna
45e71f32e8
Pass date for analysis, compute age correctly
2024-04-15 15:21:35 -04:00
Vecna
1a8b64080d
Refactor Analyzer trait
2024-04-15 13:19:56 -04:00
Vecna