I haven’t been updating these last two weeks because I actually ended up getting Covid. Turns out my tracker couldn’t give me the heads up….
The last week I ran predications for was 12/21, Christmas week. And I did pretty well! My prediction accuracy for the whole week was 93.38%!!!
Bringing my total overall accuracy up to 83.72%
TRAVIS COUNTY (AUSTIN, TEXAS)
12/21 Prediction: 3000
Actual: 2932
Accuracy: 97.73%
12/28 Actual: 3532
I noticed that the 4WASR (4 Week Average Spread Rate) forecasting formula has been pretty close the last few weeks.
Week: (Prediction | Actual)
12/28: 3448 | 3532
12/21: 3528 | 2932 ***Prediction with my “Holiday Math” comes to 2520***
12/14: 2523 | 3051
12/7: 2712 | 2232
I’ll be looking at the 4WASR forecast for the coming weeks.
FORMULA FORECASTS
4 Week: 1.14 Spread Rate = 4030 Cases
3 Week: 1.18 Spread Rate = 4159 Cases
2 Week: 1.08 Spread Rate = 3825 Cases
14 Day Average: 497 Avg = 3481 Cases
7 Days Average: 505 Avg = 3532 Cases
Those weekly averages seem to be predicting a very specific cluster of cases. While the 14 and 7 day averages are only off by 51 cases.
MY PREDICTION
The week of 12/21 only recorded Covid cases for 5 of the 7 days (71%). If we calculate what 2932 is 71% of, we get 4130.
The week of 12/28 only recorded Covid cases for 6 of the 7 days (86%). If we calculate what 3532 is 86% of, we get 4107
December saw a climb in cases almost every week.
12/1- 2164 12/7-2232 12/14-3051 12/21-2932 (or 4130) 12/28-3532 (or 4107)
If we calculate the 4WASR but use the estimated totals instead of the Holiday affected totals, we get a Spread Rate of 1.19 for the week of 1/4/21. Against the Estimated 4107 total from 12/28, that means the 4WASR estimate for 1/4/21 is 4873.
I will agree with this number. I think we will see a spike in cases due to Holiday gatherings. I’d even argue the number of cases could be higher than 4873. This spike could last two or three weeks before finally leveling out. Hopefully, the lack of any meaningful Holidays coming up (Valentine’s Day is fake) means the Spread Rate will start to level off and trend downwards.
Prediction: 4900 cases for 1/5
PASSAIC COUNTY (NEW JERSEY)
12/21 Prediction: 1500
Actual: 1714
Accuracy: 87.51%
12/28 Actual: 1428
Passaic reported cases on all days, including Christmas Eve/Day and New Years.
FORMULA FORECASTS
4 Week: 0.84 Spread Rate = 1194 Cases
3 Week: 0.82 Spread Rate = 1171 Cases
2 Week: 0.82 Spread Rate = 1172 Cases
14 Day Average: 224 Avg = 1571 Cases
7 Days Average: 204 Avg = 1428 Cases
The weekly averages being lower than the daily averages may be hinting at a spike coming with the Holidays.
MY PREDICTION
Passaic recorded cases on all days, so I won’t have to do any creative accounting.
However, that doesn’t mean that people getting tested was lower due to the Holidays. I got sick on Christmas Eve and waited until Saturday to get tested. So numbers can still be low. Not to mention I have suspected there will be a delayed boom as people travel and gather for the holidays.
The weeks leading to the holidays saw case increases like: (12/1-12/14) 2927, 2592, and 2122. Which was trending down already.
My formulas don’t support this idea but I believe there will be a spike from the Holidays.
My prediction: 2500 cases
SUFFOLK COUNTY (BOSTON, MASSACHUSSETTS)
12/21 Prediction: 3361
Actual: 3425
Accuracy: 98.13%
12/28 Actual: 3846
FORMULA FORECASTS
4 Week: 1.01 Spread Rate = 3875 Cases
3 Week: 0.99 Spread Rate = 3825 Cases
2 Week: 0.99 Spread Rate = 3797 Cases
14 Day Average: 553 Avg = 3872 Cases
7 Days Average: 609 Avg = 4262 Cases
MY PREDICTION
Both 12/21 & 12/28 were both missing one day of reporting.
So, following the same formulas… The estimated actuals would be:
12/28: 4487
12/21: 3996
Adding this to the Weekly Case counts in December we get: (11/30-12/28) 3810, 3985, 4023, 3996, 4487. So we can see a steady climb. The three weeks prior to the start of December were all hovering around 2000.
Using those numbers makes the current 4WASR forecast 1.05, 1.05 of the estimated 4487 = 4690.
But, staying in line with my other areas and my prediction of a spike, I believe this number may be closer to 5000.
Prediction: 5000 cases
RHODE ISLAND (Yea, like the whole state)
12/21 Prediction: 4624
Actual: 5129
Accuracy: 90.15%
12/28 Actual: 7134
FORMULA FORECASTS
4 Week: 0.98 Spread Rate = 6960 Cases
3 Week: 0.98 Spread Rate = 6997 Cases
2 Week: 1.10 Spread Rate = 7869 Cases
14 Day Average: 876 Avg = 6132 Cases
7 Days Average: 1019 Avg = 7134 Cases
MY PREDICTION
Rhode Island reported cases for all holidays, so I won’t have to do any creative accounting.
Since Rhode Island introduced “The Pause” a few weeks ago, they saw cases go from 8903 & 8545 a week to 6293 then 5129. A significant decrease. Yet, on New Year’s Eve and the two days prior, the state saw daily cases of 1187, 1160 and 1592. The week of 12/28 had a total case count of 7134. So we can see the spike is already starting to happen.
I think Rhode Island will see numbers next week that are close to pre-“pause” numbers.
Looking at historical spikes and their spread rates we have the Week before Halloween (1.72), the Week after Halloween (1.27) and the week after Thanksgiving (1.50). Using these Spread Rates against 12/28s cases we get:
1.72= 12,244 Cases
1.27= 9073 Cases
1.50= 10,713 Cases
Also to be considered, I’ve been looking at the states Weekly Positive Test Averages.
This chart represents the Percent of Positive Cases compared to the Total amount of tests.
Overall I think we could be looking at anything between a 1.27 and 1.50 Spread Rate. But we also haven’t seen weekly case numbers surpass 8902 before…..so…..who knows.
Prediction: 9000 Cases