Last weeks total Prediction Accuracy was 91.33%
My total accuracy rating for this little project is now 85.29%.... Almost 1% higher than last week.
TRAVIS COUNTY (AUSTIN, TEXAS)
Prediction: 5000
Actual: 4912
Accuracy: 98.24%
Spread Rate for 1/11: 1.25
The county didn’t report any new cases on the 12th. However, there was a spike in case counts for the next two days, I assume that spike was actually all the cases that came in on the 12th and went unreported. Or maybe the county didn’t do testing on that day and there was an influx of tests being done on the 13th and 14th. Either way doesn’t matter, my prediction was right….and I’m oh so humble about it.
Something to be noted is that last week’s case increase was the highest case increase in a single week since 4th of July week.
FORMULA FORECASTS
4 Week: 1.13 Spread Rate = 5561 Cases
3 Week: 1.19 Spread Rate = 5842 Cases
2 Week: 1.18 Spread Rate = 5804 Cases
14 Day Average: 630.43 Avg = 4413 Cases
7 Days Average: 702 Avg = 4912 Cases
Last week the closest prediction was the 4WASR. It predicted 4541 cases compared to the actual 4912, a 7.55% error.
It should also be noted that the 7 & 14 Day averages have been rising steadily for the last couple weeks. They are both at their all-time highs.
MY PREDICTION
Last week’s Spread Rate was 1.25. As discussed in last week’s post, Austin saw a 3 week spike after Halloween and Thanksgiving. And I suspect that the double holiday we just celebrated could cause a slightly elongated spike this time around, say 4 weeks. 1/18 is the third week.
If this week sees another 1.25 Spread rate we would get: 6164 cases
That number is actually pretty close to our 3 & 2 WASR formulas. So maybe I will stick with my 3WASR prediction.
Prediction: 5840
PASSAIC COUNTY (NEW JERSEY)
Prediction: 2000
Actual: 2016
Accuracy: 99.21%
Spread Rate for 1/11: 0.41
Last week’s case count was 4886 which was a huge spike from previous weeks. That Monday (1/4) had 2973 cases alone, so I knew it had something to do with a reporting and testing result delays. The week prior was only 1428 cases, but I knew we were going to be looking at a natural spike, which is why I predicted 2000.
Backing up my claim that those 2973 cases are more of a result/data dump on a single day…. Our lowest spread rate in the last few months was 0.81. If those 2973 cases were real cases then last week should have seen far more than 2016 new cases and a spread rate of 0.41.
FORMULA FORECASTS
4 Week: 1.37 Spread Rate = 2753 Cases
3 Week: 1.55 Spread Rate = 3128 Cases
2 Week: 1.91 Spread Rate = 3852 Cases
14 Day Average: 492 Avg = 3441 Cases
7 Days Average: 288 Avg = 2016 Cases
Just like last week, my forecasting formulas are going to be way off thanks to the 4886 cases from Jan 4th.
MY PREDICTION
As I said before, the formulas are skewed from that test result spike.
Looking at Halloween and Thanksgiving, Passaic County only had a spike lasting 2 weeks both times. So I think Christmas/New Years will see a 3 week spike. Meaning we are approaching our last week of this proposed spike.
Purely looking at the last two weeks, and factoring out those 2973 cases in a single day, we can see the weekly numbers landing around 2000. My guess would be that next week will land around that number again. Maybe even climb shortly.
Prediction: 2000….again
SUFFOLK COUNTY (BOSTON, MASSACHUSSETTS)
Prediction: 6000
Actual: 4220
Accuracy: 70.33%
Spread Rate for 1/11: 0.80
Boston was my worst prediction last week. I predicted a spike and we actually got a drop. So maybe we will see the case counts continue to drop or even level out. Last week saw 1033 less cases than the previous week, this is promising!
FORMULA FORECASTS
4 Week: 1.04 Spread Rate = 4371 Cases
3 Week: 1.10 Spread Rate = 4631 Cases
2 Week: 1.08 Spread Rate = 4577 Cases
14 Day Average: 677 Avg = 4737 Cases
7 Days Average: 603 Avg = 4220 Cases
All formulas seem to point at a slight tick upward, yet stay near last week’s numbers.
MY PREDICTION
So Boston pulled a fast one and made a nice recovery. However, historical Holiday spikes tell me that holidays typically effect 3 weeks. So I’m actually going to side with my formulas here and say there will be a slight tick upward next week. But overall, my guess is that these numbers will start to decline over the next few months.
Prediction: 4600
RHODE ISLAND (Yea, like the whole state)
Prediction: 7000
Actual: 6829
Accuracy: 97.56%
Spread Rate for 1/11: 0.96
I was 2.44% off from predicting an entire states Covid Case increase….
Also, last week Rhode Island officially surpassed 100,000 cases. That’s 1/10th of its population.
FORMULA FORECASTS
4 Week: 1.04 Spread Rate = 7108 Cases
3 Week: 1.12 Spread Rate = 7622 Cases
2 Week: 0.98 Spread Rate = 6683 Cases
14 Day Average: 998 Avg = 6987 Cases
7 Days Average: 976 Avg = 6829 Cases
Only the 4WASR formula has been notably close the last two weeks. And even that doesn’t mean much.
MY PREDICTION
Holiday spikes in Rhode Island last almost three weeks. I predict this one to last three weeks which means we are entering our last week.
That being said, last week saw 316 less cases than the week before, so while the spike results in high weekly case counts, we may actually start to see a decline/ leveling off of that spike.
Rhode Island has been a notoriously bad state with the spread rate being higher than the other 3 places I’ve been looking at. So while history tells me we should be seeing a decline coming, I almost feel like the numbers won’t dip too much.
Prediction: 6600