Name: | chimera |
---|---|
Abbreviation: | LUcompUncertLab-chimera |
Owner: | lshandross |
Team name: | LUcompUncertLab |
Description: | KM27 model trained on pct positive,HHS hosps,Human judgment |
Contributors: | Tom McAndrew (Lehigh University) <mcandrew@lehigh.edu> |
License: | Creative Commons Attribution 4.0 |
Notes: | Designated model // Data inputs: Daily flu hospitalizations,Daily percent positive cases,metaculus human judgment forecasts // Team funding: CSTE/CDC |
Citation: | |
Methods: | 2023-10-11: SEIRH model variant was fit bc KM27 model was not yet complete. // 2023-10-12: KM27 model is completed and ready for training/submission for the following week. // 2023-10-18: KM27 model was incorrect last week. i spent the entire week correcting the model. There is a final bug in the method i use to aggregate from day forecasts to week forecasts. I will need to solve this bug before the next submission date. // 2023-11-01: Corrected KM27 and added in knowledge about flu: R0 is constrained to be between 1.1 and 2.5, vaccine proportion is between 0.2 and 0.8. Generatioanl interval is Neg Bin and on average 3.6 days. / Changed from Normal observational process to GammaPoisson. The GP is good but very computationally slow. / Still cranky about how sensitive this model is to changes. This round ill try to focus on assigning seasons to distinct clusters. / 2023-11-08:Added in priors for peak intensity and timing, and for the first two weeks of the season (MMWR 40 and 41). This adds stability to the model and speed. // 2023-11-23:Found two major bugs in the code. One is frustrating and the other very subtle. First bug was training on hospitalization data the same as last year. i missed the message about a change in the hosp data and so have been training my model on the wrong data up until now. The second bug is a subtle fix in the code that took some time to find. // 2023-12-09: Added back QO data, truncated to be two weeks before the forecast submission date. Forecasts look ok except for FIP=72. i'll need to investigate. // 2023-12-16: Ran versions without and without the QO data. For the majority of locations i chose the model that included QO data. There were 3 locations, 30,37,56 where i thought the model without QO data was better and so chose those forecasts instead. // 2023-12-30: Not much to report. Still running a few slightly different versions of the model and then hand picking which model to submit for which location." // 2024-03-20:Not in love with Loc 2 but otherwise forecasts look ok" |
Home: | https://github.com/computationalUncertaintyLab |
Auxiliary data: | (No URL) |
Timezero | Data Source | Upload Time | Issued at | Version |
---|---|---|---|---|
2024-04-27 | forecast-738179ff33d0.json | 2024-04-25 11:22:32 UTC | 2024-04-25 11:22:32 UTC | |
2024-04-20 | forecast-734e4717c087.json | 2024-04-18 11:22:22 UTC | 2024-04-18 11:22:22 UTC | |
2024-04-13 | forecast-73e312e53afa.json | 2024-04-12 11:22:42 UTC | 2024-04-12 11:22:42 UTC | |
2024-04-06 | forecast-73a779e89c5d.json | 2024-04-04 11:22:22 UTC | 2024-04-04 11:22:22 UTC | |
2024-03-30 | forecast-758270e2644e.json | 2024-03-28 11:26:10 UTC | 2024-03-28 11:26:10 UTC | |
2024-03-23 | forecast-75553741acb.json | 2024-03-21 11:26:10 UTC | 2024-03-21 11:26:10 UTC | |
2024-03-16 | forecast-75a45a475158.json | 2024-03-14 11:25:02 UTC | 2024-03-14 11:25:02 UTC | |
2024-03-09 | forecast-75e26d24940f.json | 2024-03-07 11:25:20 UTC | 2024-03-07 11:25:20 UTC | |
2024-03-02 | forecast-75202aa8cde1.json | 2024-03-01 15:33:13 UTC | 2024-03-01 15:33:13 UTC | |
2024-06-01 | (No data) | |||
2024-05-25 | (No data) | |||
2024-05-18 | (No data) | |||
2024-05-11 | (No data) | |||
2024-05-04 | forecast-73ae51838356.json | 2024-05-02 11:22:46 UTC | 2024-05-02 11:22:46 UTC | |
2024-02-24 | forecast-74e74b1bf655.json | 2024-02-22 11:24:57 UTC | 2024-02-22 11:24:57 UTC | |
2024-02-17 | forecast-754721bb0db8.json | 2024-02-15 11:24:42 UTC | 2024-02-15 11:24:42 UTC | |
2024-02-10 | forecast-74e8181660a7.json | 2024-02-08 11:25:11 UTC | 2024-02-08 11:25:11 UTC | |
2024-02-03 | forecast-750e1e969b06.json | 2024-02-02 11:24:23 UTC | 2024-02-02 11:24:23 UTC | |
2024-01-27 | forecast-756e5bc82906.json | 2024-01-25 11:26:13 UTC | 2024-01-25 11:26:13 UTC | |
2024-01-20 | forecast-74f853130d7d.json | 2024-01-18 11:25:02 UTC | 2024-01-18 11:25:02 UTC | |
2024-01-13 | forecast-7520131c29f.json | 2024-01-11 11:25:08 UTC | 2024-01-11 11:25:08 UTC | |
2024-01-06 | forecast-74f9c9076f9.json | 2024-01-04 11:25:04 UTC | 2024-01-04 11:25:04 UTC | |
2023-12-30 | forecast-74df5ebf6f70.json | 2023-12-28 11:24:56 UTC | 2023-12-28 11:24:56 UTC | |
2023-12-23 | forecast-74d1689fefe8.json | 2023-12-21 11:25:12 UTC | 2023-12-21 11:25:12 UTC | |
2023-12-16 | forecast-75bc8bbfccc.json | 2023-12-14 11:25:09 UTC | 2023-12-14 11:25:09 UTC | |
2023-12-09 | forecast-7532357918a.json | 2023-12-08 14:56:39 UTC | 2023-12-08 14:56:39 UTC | |
2023-12-02 | forecast-65b01fc63a44.json | 2023-12-07 15:03:39 UTC | 2023-12-07 15:03:39 UTC | |
2023-11-25 | forecast-44845fe231df.json | 2023-11-28 16:24:53 UTC | 2023-11-28 16:24:53 UTC | |
2023-11-18 | forecast-44845d5c19f5.json | 2023-11-28 16:24:51 UTC | 2023-11-28 16:24:51 UTC | |
2023-11-11 | forecast-44845ed06157.json | 2023-11-28 16:24:49 UTC | 2023-11-28 16:24:49 UTC | |
2023-11-04 | forecast-4484669f1409.json | 2023-11-28 16:24:47 UTC | 2023-11-28 16:24:47 UTC | |
2023-10-28 | forecast-44842f0c38e8.json | 2023-11-28 16:24:45 UTC | 2023-11-28 16:24:45 UTC | |
2023-10-21 | forecast-4484bbb1cca.json | 2023-11-28 16:24:43 UTC | 2023-11-28 16:24:43 UTC | |
2023-10-14 | forecast-44842c2a5e80.json | 2023-11-28 16:24:41 UTC | 2023-11-28 16:24:41 UTC | |
2023-10-07 | (No data) |