Current total accuracy rating: 84.49%
TRAVIS COUNTY (AUSTIN, TEXAS)
Last week’s numbers didn’t hit the mark but the county also failed to report any cases on Jan 10th. Then Jan 11th cases were a bit higher than usual, so I think they just didn’t get the numbers in on time on the 10th???
My prediction was for 4900 cases, the actual was 3914. 79.87% accuracy.
Just for giggle let’s pretend the 10th got the daily average number of cases …. The daily average (1/4-1/9) was 652. So 3914 + 652 = 4566 which is 93.18% accuracy.
Not bad! Makes me wonder what happened on the 10th?
FORMULA FORECASTS
4 Week: 1.16 Spread Rate = 4541 Cases
3 Week: 1.09 Spread Rate = 4271 Cases
2 Week: 1.16 Spread Rate = 4526 Cases
14 Day Average: 531.86 Avg = 3723 Cases
7 Days Average: 559 Avg = 3914 Cases
The 2WASR forecast has actually been pretty close the last two weeks. It forecasted 3413 for 12/28, the actual was 3532. Then last week it forecasted 3825.
The 4WASR was also pretty good. It forecasted 3448 for 12/28 and 4030 for last week.
MY PREDICTION
12/28 had a 1.20 Spread rate.
1/4/21 had a 1.11 Spread Rate. If the daily average was reported on the 10th, the spread rate would be 1.29.
So while this past weeks numbers were low, I think it actually has to do with a reporting error. The case count for the 11th is 955, higher than the average, which tells me that the non-reported cases from last week will carry over into this week.
Seeing as the 4WASR and the 2WASR have been pretty good these last two weeks, I am going to look at their forecasts. But, I have to add on what I assume are last weeks missing numbers. So while the 4WASR forecast is for 4541, I’m going to go for 5000.
Prediction: 5000
Spread Rate
Weekly: Daily Case Averages
PASSAIC COUNTY (NEW JERSEY)
These numbers were double what I expected. I forecasted 2500 cases, but the actual was 4566.
However, the 4th saw 2973 new cases reported. But the daily average for the rest of the week (5th – 10th) was only 316. Replacing the super high numbers on the 4th and replacing them with the daily average gets me 2208 cases and an accuracy of 88.34%.
But that’s not the real world, so what happened?
Well I think this is the influx of cases we get from the Holidays. From personal experience I know that it takes a few days to get results from the test. I also know that human nature says, “Let’s go out and parrrrrtaaaaaay.” So I think two things happened.
First, there is a delay in results from people getting tested from the 1st to the 3rd, those results come in on the 4th and inflate that week’s numbers. Second, I’d be willing to bet there was an influx of testing and positive tests due to all the holiday shenanigans. People party with someone at Christmas or New Year’s, a few days later they find out their friend at the party was exposed and now they need to get tested. So that huge number on Monday can be people who were exposed at Christmas but didn’t know yet.
So what’s my accuracy? Well it could be 51.37% (2500/4866) or it could be 88.34% (2208/2500)
I’ll take the ‘L’ on this one, but to be fair….I was also kind of right.
FORMULA FORECASTS
4 Week: 1.47 Spread Rate = 7137 Cases
3 Week: 1.68 Spread Rate = 8189 Cases
2 Week: 2.12 Spread Rate = 10318 Cases
14 Day Average: 450 Avg = 3147 Cases
7 Days Average: 695 Avg = 4866 Cases
These numbers are all super off thanks to that crazy spike on the 4th.
MY PREDICTION
Last week’s Spread Rate was 3.41. Which we can probably label as a freak event thanks to the Holidays and maybe a backlog of test results?
For 4 weeks prior to Christmas/New Years, the Spread Rate in Passaic County has been hovering around 0.84. So each week there was a steady decline in the case increase. Next week we may see a crazy low spread rate as the case increase peaks with the Holidays. OR….it halves as now there are more infected people in the area and we see 2500ish cases instead of the sub 2000 case count the previous weeks saw.
This prediction is tough. Theres more infected people to go out and infect others. But I also don’t know the actual reason for the spike, if it’s a back log and reporting issue, then it has no effect on the real world spread rate. Regardless, I predicted the Holidays would be responsible for a spike over the next two weeks. So, ignoring all my forecasts and going with my gut here….
Prediction: 2000
Spread Rate Actuals
Passaic Actual Spread Rates Compared to Forecasts
SUFFOLK COUNTY (BOSTON, MASSACHUSSETTS)
I did pretty good in Suffolk. I predicted 5000 cases, the actual was 5253. Accuracy of 95.18%!!
Boston gave me no real surprises. The daily case count was higher than previous weeks (as predicted). And all days are accounted for. Good job guys. Go get yourselves a coffee at Dunks. On me.
FORMULA FORECASTS
4 Week: 1.09 Spread Rate = 5712 Cases
3 Week: 1.11 Spread Rate = 5849 Cases
2 Week: 1.24 Spread Rate = 6537 Cases
14 Day Average: 680 Avg = 4758 Cases
7 Days Average: 800 Avg = 5603 Cases
I should probably note that none of the formulas have been notably close to hitting the right numbers in the past few weeks.
MY PREDICTION
Last week’s Spread Rate was 1.37, the previous week was 1.12 and the weeks before that sat at around 1.01ish. Just as I thought, a spike around the Holidays. But what now? When does the spike flatten out and start to dip? Well with no upcoming Holidays, I predict things to head back down soon. But I could see this week’s numbers looking close to last weeks as Boston begins to flatten out. So maybe the 4WASR and 3WASR formulas have a point?
Harking back to Halloween and Thanksgiving, both Holidays saw a two week spike and then slope back to Spread Rates of 0.95-1.00. Seeing as Christmas and New Year’s kind of sit on top of each other, I think all states/cities may see a three-plus week spike. Kind of like compounding interest on a bad loan.
All the formulas are close to the 6000 mark. And given the pattern of past holidays I’d be willing to say that’s a good mark to aim for.
Prediction: 6000
RHODE ISLAND (Yea, like the whole state)
Rhode Island also didn’t exactly hit my number, but it didn’t entirely disappoint. 5 out of 7 days last week saw over 1000 new cases each.
My prediction was for 9000, the actual was 7145. 79.39% accuracy.
The state reported cases on all days and no day was any more unusual than another.
Also, as of January 11th, the state has surpassed 100,000 Covid Cases reported. That’s one-tenth the state’s population. Meaning if you know 10 people in Rhode Island, odds are that you know one person who has had Covid.
Also also, I’ve been following the Case Count Doubling timeline.
Apr 10: Cases Doubled in 5 days
Apr 19: Cases Doubled in 9 days
May 6: Cases Doubled in 17 days (10638 cases)
Aug 24: Cases Doubled in 110 days
Nov 14: Cases Doubled in 82 days
Dec 28: Cases Doubled in 44 days (85601 cases)
The next predicted Doubling is in roughly 80 days. But that’s only if there are 7000+ cases a week for the next three months. Unless I’m looking at this wrong…It looks like the doubling may have peaked in December. If the cases really do start to slope down as I predict, there won’t be another Doubling for over 120 days.
FORMULA FORECASTS
4 Week: 0.99 Spread Rate = 7045 Cases
3 Week: 1.07 Spread Rate = 7639 Cases
2 Week: 1.20 Spread Rate = 8547 Cases
14 Day Average: 1020 Avg = 7140 Cases
7 Days Average: 1021 Avg = 7145 Cases
Again, no formula has been notably accurate in the last few weeks.
MY PREDICTION
Looking at Halloween and Thanksgiving again…Halloween had a 3 week Spread Rate spike and Thanksgiving had 2ish weeks. Again, Christmas and New Year’s being so close together may spark a 3 week spike. But, last week’s spike was not as big as I expected. Last week’s Spread Rate was almost exactly 1.00. So maybe we will see another 1.00 week before a slow creep downward. OR….. Rhode Island goes crazy next week and everyone infects 5 people each….
Prediction: 7000
BONUS!!!!
I’ve been looking at Rhode Islands Testing results as well. Specifically tracking the Positive % compared to total tests for the week. Take a look at this chart.
