Gained a 12% accuracy increase for my problem child category.
Data Cleaning, 800 lines of new training data and 3 additional training runs.
Next I’m going to try introducing custom stop words in an attempt to further the cleaning process.
Update:
Adding the cleaning process decreased the cumulative score of both models by 2 points. I’ll turn that off for now, but I think training a model from scratch, with the cleaner turned on, may actually be the better idea.
I have 18 categories to classify.
Each category contains anywhere from 3 - 140+ classes.
15 of my categories are routinely scoring 90% and above when back tested. 2 others are scoring between 80-90% and the last is my problem child. The last category is my largest with 140+ classes in it for the model to choose from. This category started at 25%.
PROBLEM 1: DIRTY DATA
One of the first things I noticed were mistakes in the classification for this class. Misspellings or just straight up incorrect class choice. One of the most common misspelling was making a singular item plural.
To combat the multiple potential problems, I created a script to go through every class of each category and compare it to the other classes within it’s own category. It would then return a “similarity score” and return any classes that were over 80% similar. This uncovered all the misspelled and pluralized classes to which I then updated within the dataset.
Fixing these increased accuracy by 5-10%.
PROBLEM 2: VARIANCE
My next problem was dealing with variance in similar classes within one category. For example: 1/4in Hex Bolt is similar to 1/2in Hex Bolt, yet they are two completely different items.
For a time I was attempting to classify both within the same category. With 30 1/4in examples and 42 1/2in examples, compared to other classes with 100+ examples, the model struggled to identify either. That’s when I decided to create a “Sizing” category. Hex Bolt is the “Product” and 1/4in or 1/2in is the size. So now the model has 70 examples of Hex Bolt as a “Product”. And since “Sizing” is only dealing with two classes, it has an easier time identifying them.
I continued this for all other classes that could fit this scenario. Consolidating similar items and separating the various characteristics that I could.
This increased the “Product” accuracy by 15% while “Sizing” came in above 90%.
Now my problem child is coming in at 45% single shot training and nearly 50% on second training.
I think the next thing I need to focus on is the parameters for training.
I’ve already consolidated as much as I can. The only other idea I have is to breakout different models for different item types. Such as all the Bronze items have their own model, Silver has their own. But the computing power for the actual classifying is limited. Having to run 5+ models per classification will take over an hour to do. I want to perform this quicker.
I first had the idea for this particular project about three years ago. I'm now a year into the project and I have learned much. Yet I feel like I know nothing.
What is an NLP?
A machine learning technology that enables computers to understand, process, and manipulate human language. NLP is a branch of artificial intelligence, computer science, and linguistics. It uses techniques like machine learning, neural networks, and text mining to interpret language, translate between languages, and recognize patterns. NLP is used in many everyday products and services, including search engines, chatbots, voice-activated digital assistants, and translation apps.
Why an NLP?
I had tried several approaches and tested a few ideas. Ultimately I realized that my raw inputs would be too messy for most models. I needed something flexible and capable of comparing words within a sentence. NLP seemed like the best option that required the least physical resources.
What model am I using?
I am currently using a pre-trained Distilled BERT model that I am fine-tuning with my custom data.
How do I train it?
I will probably get into this with more detail at a later date, mostly because I want to update this part. However, I am converting the training data into a DataFrame with Python and then splitting that data into train and validation sets. But I feel like I can improve this significantly.
Goal
Classify an incredibly large dataset with at least 85% accuracy on an hourly basis without human assistance.
Problems
There are no training datasets for public consumption. I need to create my own.
The dataset needs to be fairly big in order to get the best results.
The custom dataset needs to be classified manually, which takes longer the bigger it is.
There will be over 360 different classes in the dataset. There needs to be a balance.
My processing power is limited to a 6 year old GTX1070
The first couple problems have already been solved/still being solved. I have created 36,000+ lines of training data by scraping data that gets sent directly to me on a daily basis. And also, since I’m a data hoarder, I still have three years worth of raw data to convert into usable training data.
The third problem is still an ongoing problem. It takes a long long time to classify 36,000+ lines of training data. And my plan for the summer is to have 40,000 lines of training data. My next problem is that while I gain good training data for some classes, I’m still lacking for other classes. So I have to hunt for training points on lesser utilized classes. And they are lesser utilized for a reason. This slows down the overall progress of the project as it takes time to claw for examples of these points.
CUE THE PONZI SCHEME
This is when I came up with an idea. The NLP model reads the raw data and makes a classification effort for 18 different categories. Each category could relate to anywhere from 1 to 5 words in the raw data. Those words can be spelt or abbreviated different several ways that all mean the same thing. By swapping the source words with their alternatives, I can inflate the training data. And depending on the amount of alternative words in a single source sentence, that sentence can be transformed as many as 15 times. Now the model not only gets reinforcement training, but exposure to all spelling and abbreviation types.
That approach turned my 36,000 lines into 293,000.
NEXT
So now I need to ponder my processing power problem. My GTX 1070 doesn’t do a terrible job. But the bigger the model gets, the longer it takes to train. A few ways I think I can approach this without buying hardware:
Adjust Training Parameters
Play with the padding / truncation
Clean up the training data
Consolidate the Category with 140+ possible classes
Research
So I’m thinking about moving to Rhode Island and wanted to be able to have some sort of idea how the markets are up there. When is the best time to buy? How is the market growing? Or is it shrinking? How quickly do I need to make a move?
To answer these questions I turned to Python.
A little research, and I really do mean “little”, brought me to Redfin. Redfin has publicly accessible data covering varying topics all across the United States housing market.
The best part is that I can load this data directly into Python without having to download a CSV or Zip File.
url = 'https://redfin-public-data.s3.us-west-2.amazonaws.com/redfin_market_tracker/county_market_tracker.tsv000.gz' National_Data = pd.read_csv(url,compression = 'gzip', sep ='\t', on_bad_lines = 'skip')
That dataset contains market data for counties all across the US since 2012. Given I only care about Rhode Island and any data prior to Covid would be useless, I narrowed the data down to everything after 2021 in the following Rhode Island counties:
Bristol County
Kent County
Newport County
Providence County
From here I have quite a few pieces of data to look at since Redfin tracks 30+ categories per monthly entry. Some categories include Number of New Listings, Pending Listings, Sold, Percent of the market whose price dropped, Number of homes sold above listing, Number of listings off the market within two weeks, Median list price, median sold price, the monthly and yearly averages for those numbers and so on.
So what do I care about? Great question. I’m not quite sure that I know what I’m looking for but I know what interests me and what might give me some ideas. So I created a series of graphs.
What does this graph tell me? Well it gives me an idea of how fast the market is moving and its relative demand. If New listings outpace Pending and/or Sold, then it’s a buyers market. Alternatively, if New can;t keep up with Peding/Sold, it’s a sellers market. And if Sold Above List price goes up, then demand is even higher.
Here I can figure out if the Median List Price has some wiggle to it. If the Median Sale price is higher than the list price, there may be more demand in the area. Or the listed price is just a conversation starter. However, if the Median Sale price is lower than the List price, that means the market may be softening and the buyer has more negotiating power. Which would be good to know for the county you are buying in.
This graph is similar to the first graph in that it tells me the markets strength at current pricing. Are the homes selling for more or less than originally listed for? And how quickly are they moving off the market? Is it above listing and off the market in two weeks? Or is the price dropping and staying on the market?
Finally, in order to make this report useful to me, I automated it to pull at the beginning of each month and to email me personally when it does. I get a curated email for each county giving me insight into how that market is performing and how I might want to enter it.
I do kind of miss the ramblings of a madman that I used to do here while reporting and predicting Covid case counts.
I need a space to type out my thoughts while I work on these bigger projects, sans alcohol.
Currently I am monitoring the success rate of a few stock forecasting scripts. Since I started using Python to inform me of good stocks three months ago, I have seen a 27% return. I’ll probably get into more detail on the subject later, but it includes using Python to do a few things.
Read over the last decade+ of data that Yahoo Finance has on the give stocks and create a ML model to predict the next business days closing price. It’s not usually very accurate on the price, but it is fairly good at predicting the direction the stock will go.
Use MACD, RSI, DMA, SMA30, SMA14, SMA30 together to identify buy and sell signals on a daily basis and alert me when the stock enters either signal.
I then get these reports every morning via email.
I’m looking towards improving these approaches and other signals I should look towards.
This twitter account is powered entirely by automatic bots that I created. They comb through several data points that I find interesting and post about them on a weekly schedule. This includes the protein, housing and stock markets. As well as tracking COVID cases and doing some light tweet sentiment analysis.
This allows me to keep an eye on several topics without having to take the time to manually track or update formulas.
Seeing all the News outlets talk about the ‘sudden’ spike in Covid Cases and the new Variant running around, I got curious and wanted to see where we were last year at this time.
The following charts compare NEW cases on the same dates (and future dates) from 2020 to 2021.
Dotted lines are 2020 to 2021 (March to February)
Solid lines are 2021+ (March and on)
Texas, Massachusetts, New Jersey and Rhode Island
Massachusetts only
New Jersey only
Rhode Island only
Texas only
MY PREDICTION
Austin hit lower than I thought. But I don’t think it will continue to drop. This past week saw a lot more social interaction. The new soccer team drew large crowds at multiple viewing parties across the city. The Night Life on Saturday was very active on 6th street as well. My guess is that the cases will at least stay in the area they are now, if not bump up a little.
Prediction: 800
MY PREDICTION
Passaic cases hit higher than I thought but is continuing to climb at a similar pace to the weeks prior. That pace lines up with the 4 & 3WASR forecasts.
Prediction: 2300
MY PREDICTION
Boston is weird. The case count from last week actually falls in line with what I would have thought the cases would be three weeks ago. But 3/29 was when the spike started and threw my predictions off. But all my forecasts stay around the 1.0 Spread Rate zone.
Prediction: 1500
MY PREDICTION
This state was close enough to my prediction but still hit high, again. I’ll forego the “Untouched” formulas as they obviously need some work. Looking at my WASR formulas, it looks like the 3 Week Avg forecast has been the closest the last two weeks.
Prediction: 3200
UPDATE: I finally got around to automating my Covid Reports. Now all my Weekly Round Ups and and Forecasting Formulas will be presented in form of a screenshot of my spreadsheet and not actually typed out. Probably saves me an hour each week of shifting between screens.
Last weeks accuracy was: 77.33%
Total Project Accuracy is: 84.09%
TRAVIS COUNTY (AUSTIN, TEXAS)
MY PREDICTION
Cases are going to spike as people get braver with going into public spaces and fewer places require masks. I don’t see this turning into a ‘run away’ spike like the holidays were. More vaccinations limit the amount of people that can get the virus and the increase in good weather and higher temps will result in more people outside. As we’ve seen before, going out and higher temps increase the probability of the Spread Rates lowering.
Prediction: 1000
PASSAIC COUNTY (NEW JERSEY)
MY PREDICTION
Passaic actually reversed on me! Given its steady rise over the last few weeks, I can’t see this number dropping by much. I’ll stick with the Forecasting Formulas.
Prediction: 1700
SUFFOLK COUNTY (BOSTON, MASSACHUSSETTS)
MY PREDICTION
Boston was my closest prediction but its still higher than I expected. The Spread Rate Forecasters point towards a continued rise.
Prediction: 2300
RHODE ISLAND (Yea, like the whole state)
1WUSP (1 Week Untouched Spread Predictor): 1270
2WUSP: 1309
3WUSP: 1419
MY PREDICTION
Rhode Island continuing its rise. WASR Formulas point towards a nearly 1.0 Spread Rate. But WUSP Formulas point towards almost half that. Then again, the WUSP operates under the assumption that Vaccine Rates continue to rise.
Prediction: 2900
I skipped 3/29 due to a career change.
3/22 Accuracy Rating: 87.35%
Total Project Accuracy Rating: 85.17%
Prediction 3/22 : 650
Actual 3/22: 714
Accuracy: 91.04%
Spread Rate for 3/22: 1.01
Actual for 3/29: 586
Spread Rate for 3/29: 0.82
FORMULA FORECASTS
4 Week: 0.88 Spread Rate = 514 Cases
3 Week: 0.89 Spread Rate = 521 Cases
2 Week: 0.91 Spread Rate = 536 Cases
14 Day Average: 93 Avg = 650 Cases
7 Days Average: 84 Avg = 586 Cases
MY PREDICTION
All formulas are pointing at a low to mid 500s. Monday (4/5) had a higher than usual count thanks to no case reports on 4/4, Easter Sunday. Given the Holiday and increased Social interactions, I would guess the numbers this week will not dip too much. Potentially even result in a 1.0 Spread Rate, or close to it.
Prediction: 580
Prediction: 1950
Actual: 1885
Accuracy: 96.67%
Spread Rate for 3/22: 1.07
Actual for 3/29: 2089
Spread Rate for 3/29: 1.11
FORMULA FORECASTS
4 Week: 1.09 Spread Rate = 2274 Cases
3 Week: 1.10 Spread Rate = 2301 Cases
2 Week: 1.09 Spread Rate = 2272 Cases
14 Day Average: 284 Avg = 1987 Cases
7 Days Average: 298 Avg = 2089 Cases
MY PREDICTION
Passaic just keeps climbing. The 2WASR has been very accurate lately. And all formulas are landing in the same area. I’ll go with the formulas on this one.
Prediction: 2275
Prediction: 1300
Actual: 1757
Accuracy: 73.99%
Spread Rate for 3/22: 1.33
Actual for 3/29: 1774
Spread Rate for 3/29: 1.01
FORMULA FORECASTS
4 Week: 1.08 Spread Rate = 1920 Cases
3 Week: 1.12 Spread Rate = 1990 Cases
2 Week: 1.17 Spread Rate = 2077 Cases
14 Day Average: 252 Avg = 1766 Cases
7 Days Average: 253 Avg = 1774 Cases
MY PREDICTION
Boston got a nice spike in cases two weeks ago. On top of that, the 7 Day Average has been higher than the 14 Day Average for the last three weeks. This tells me we are in a climb. Given the holiday and people getting bolder with the Vaccine roll out, I would guess the cases continue to climb. I’ll go with the 4WASR.
Prediction: 1900
Prediction: 2400
Actual: 2736
Accuracy: 87.72%
Spread Rate for 3/22: 1.08
Actual for 3/29: 2480
Spread Rate for 3/29: 0.91
FORMULA FORECASTS
4 Week: 1.05 Spread Rate = 2600 Cases
3 Week: 1.15 Spread Rate = 2859 Cases
2 Week: 0.99 Spread Rate = 2461 Cases
14 Day Average: 522 Avg = 3654 Cases
7 Days Average: 496 Avg = 3472 Cases
1WUSP (1 Week Untouched Spread Predictor): 2480
2WUSP: 2261
3WUSP: 2028
MY PREDICTION
Spread Rate Average Formulas are betting on the Spread rising. While the “Untouched” Spread Formulas are betting on the Vaccination numbers to increase, limiting the amount of people that can actually get Covid.
I’m going to side with the Vaccine logic seeing as the vaccinations have been high the past three weeks.
Prediction: 2400
Last weeks Prediction Accuracy: 85.92%
Total Project Accuracy: 85.06%
Prediction: 950
Actual: 708
Accuracy: 74.53%
Spread Rate for 3/15: 0.84
FORMULA FORECASTS
2 Week: 0.84 Spread Rate = 595 Cases
14 Day Average: 111 Avg = 778 Cases
7 Days Average: 101 Avg = 708 Cases
NOTE: Data for 3 & 4 Week Averages are still affected by the winter storm disruption.
MY PREDICTION
Cases are on a decline. I should note that Austin has stricter Mask laws than the rest of the state and that the weather has been really nice the last few weeks. Streets are filled with people going out. As I’ve stated before, the Spread Rate goes down as the Temps go up. Average Temp next week is forecasted to rise, so cases should drop again.
Prediction: 650 (Not quite a 0.84 Spread Rate, but close)
Prediction: 1650
Actual: 1766
Accuracy: 93.43%
Spread Rate for 3/15: 1.13
Looks like I was right about the cases rising.
FORMULA FORECASTS
4 Week: 1.11 Spread Rate = 1964 Cases
3 Week: 1.14 Spread Rate = 2013 Cases
2 Week: 1.09 Spread Rate = 1923 Cases
14 Day Average: 238 Avg = 1666 Cases
7 Days Average: 252 Avg = 1766 Cases
MY PREDICTION
The average rise in Spread Rates over the last four weeks is 11.216%.
Using 11.216% we translate that to a spread rate of 111.21. A Spread Rate of 111.21 for last week’s 1766 cases comes out to roughly 1960 cases. My top three forecasts seem to agree with this number.
New Jersey Cases have been rising and the Governor says they may slow down the lifting of restrictions. Code for: Do nothing. Neither lift nor restrict.
I’ll trust the math on this one.
Prediction: 1950
Prediction: 1200
Actual: 1319
Accuracy: 90.98%
Spread Rate for 3/15: 1.02
FORMULA FORECASTS
4 Week: 1.0 Spread Rate = 1315 Cases
3 Week: 0.96 Spread Rate = 1268 Cases
2 Week: 0.99 Spread Rate = 1311 Cases
14 Day Average: 186 Avg = 1304 Cases
7 Days Average: 188 Avg = 1319 Cases
MY PREDICTION
This ones tough. Mass is rolling back restrictions, they just entered Stage 4 and are allowing outdoor dining. As I have seen in Austin, outdoor activity has had a positive effect on lowering Covid numbers.
Also, cases have been hovering in the 1300 area for the last 3 weeks. Just to be safe I’ll hit a middle number. Cases can e
Also, cases have been hovering in the 1300 area for the last 3 weeks. Just to be safe I’ll hit a middle number. Cases can either go down with the increased outdoor activity or go up as socialization may rise.
Prediction: 1300
Prediction: 2150
Actual: 2537 (1880 when accounting for weekend overflow)
Accuracy: 84.75% (87.44% when accounting for weekend overflow)
Spread Rate for 3/15: 1.47
NOTE: So my previous source of Rhode Island Covid cases (The Atlantic) has stopped reporting cases. I will be moving back to my previous source: Google/NYT. The difference between the sources is that The Atlantic reported cases on the weekends while Google/NYT only reports Monday through Friday. This causes Monday to get Sat & Sun counts to be added on. This causes a slight disruption in how the data gets analyzed and forecasted.
From this point forward, I will be forecasting based on the numbers I get, but will keep in mind that the weeks numbers are not fully accurate.
FORMULA FORECASTS
Forecasts don’t work right now with the change in data gathering
MY PREDICTION
If the last two weeks had the same process of gathering data then 3/8 would have had 2381 cases. That makes last week’s Spread Rate 1.06. This is a rise from 3/1 2345 cases. Going back even further with the same process, the previous two weeks would have both been around 2400ish.
Rhode Island doesn’t plan on enacting any restrictions. Going back to my previous deep-dive into the Rhode Island Timeline of restrictions, we saw that any reduction and easing of restrictions actually had an effect on lowering cases.
That being said, the cases have been hovering in the 2400-2500 area for a while.
Side note: my “Untouched” Forecaster predicted a higher case count than it did last week. So maybe we are looking at a rise in cases?
Prediction: 2400
Last weeks accuracy was: 81.64%
Project Accuracy is now: 84.97%
Prediction: 1350
Actual: 847
Accuracy: 62.74%
Spread Rate for 3/8: 0.85
Actual numbers hit lower than expected. But I think this is because the weather has been nice and people have been outside more. I can attest to this first hand.
FORMULA FORECASTS
2 Week: 0.70 Spread Rate = 590 Cases
14 Day Average: 132 Avg = 925 Cases
7 Days Average: 121 Avg = 847 Cases
MY PREDICTION
I have no idea. Before the ice storm the cases were 1707, then we got a slight spike as two weeks of reporting rolled into one and 2/22 reported 1827 cases. Then the drop to 1002 cases and another drop to 847 cases.
So I think that Storm, like I predicted, significantly halted Covid spread. But I was wrong when I predicted the reopening measures increasing the case numbers. I predicted it too early. AS we know, there is a lag time with symptoms and a lag time with test results. So if there was any spreading this week we probably wouldn’t start to see it until next week.
Prediction: 950
Prediction: 1450
Actual: 1565
Accuracy: 92.65%
Spread Rate for 3/8: 1.05
FORMULA FORECASTS
4 Week: 1.01 Spread Rate = 1583 Cases
3 Week: 1.11 Spread Rate = 1732 Cases
2 Week: 1.15 Spread Rate = 1793 Cases
14 Day Average: 218 Avg = 1528 Cases
7 Days Average: 224 Avg = 1565 Cases
MY PREDICTION
Passaic cases have been gradually ticking upward, spread is up 400 cases since 2/15. The average increase is around 133 a week. By that math we could see cases next week hit 1698. My guess is the Spread Rate will sit between the 4WASR prediction of 1.01 and the 3WASR of 1.11.
Prediction: 1650
Prediction: 1150
Actual: 1289
Accuracy: 82.22%
Spread Rate for 3/8: 0.96
FORMULA FORECASTS
4 Week: 0.95 Spread Rate = 1220 Cases
3 Week: 0.99 Spread Rate = 1273 Cases
2 Week: 0.93 Spread Rate = 1200 Cases
14 Day Average: 188 Avg = 1313 Cases
7 Days Average: 184 Avg = 1289 Cases
MY PREDICTION
Last week my 3WASR and 2WASR were the closest to the actual case count. The week before that my 3WASR was close too. I’ll stay in the area….
Prediction: 1200
Once again, RI is the problem child of data. There are unreported days (13 & 14) the daily average was 340 a day.
I’ll have to update the numbers later this week to see what happened.
Prediction: 2100
Actual: 1721 (2361 with continued daily average)
Accuracy: 81.95% (88.95% with continued daily average)
Spread Rate for 3/8: 0.73 (1.01 with continued daily average)
FORMULA FORECASTS
Since the week was missing some reports, I’ll skip the weekly forecasting and show you something else I’m working on.
The “Untouched” helper.
RIs population is roughly 1.009Million. So far RI has vaccinated 374,193 people. There has been a total of 132,366 Covid Cases. Which means that out of 1,009,904 people- 506,559 people are potentially unable to contract Covid. (Yes I know, those who had it can get it again, and there’s a new strain and all that stuff. But this isn’t an exact science. There could also be old case + new vaccine overlap. Stick with me here.)
1,009,904 – 506,559 = 503,345 People who are capable of contracting the virus.
So what I have been doing is using a running tally of all those “Touched” by Vaccinations and the virus, subtracting them from Gen Pop and getting an “Untouched” estimate. From there I can calculate, based of prior week’s Virus Spread, what a next week’s Covid Cases could be. Keep in mind this isn’t fully fleshed out and I’m working on a worse case scenario.
1WUSP (1 Week Untouched Spread Predictor): 1533
2WUSP: 1703
3WUSP: 1771
Basically, these are saying that based on the current Covid Case increases and the Vaccination Rates, those numbers are the potential new Covid cases.
MY PREDICTION
Without those last two days of reported cases, its hard to get reliable predictions. I have predictions as low as 1533 and as high as 2418. (When accounting with and without the daily averages to fill in the missing days)
I’ll update the numbers later this week, but based on the trend I’ve been seeing….
Prediction: 2150
Last week’s accuracy rating was 82.93.
Prediction: 1600
Actual: 1002
Accuracy: 62.63%
Spread Rate for 3/1: 0.55
FORMULA FORECASTS
Week Forecasts are still all off due to the storm.
14 Day Average: 202 Avg = 1415 Cases
7 Days Average: 143 Avg = 1002 Cases
MY PREDICTION
Quick:
No Masks in Texas anymore. Austin will probably stick with masks for a little while longer.
Looks like we are starting to see the effect of that ice storm and cold weather. Preventing people from going out entirely stops the spread.
No masks + a two week waiting period from the storm = Small Spike.
Prediction: 1350
Prediction: 1100
Actual: 1491
Accuracy: 73.78%
Spread Rate for 3/1: 1.24
FORMULA FORECASTS
4 Week: 1.00 Spread Rate = 1491 Cases
3 Week: 1.00 Spread Rate = 1490 Cases
2 Week: 1.14 Spread Rate = 1693 Cases
14 Day Average: 192 Avg = 1346 Cases
7 Days Average: 213 Avg = 1491 Cases
MY PREDICTION
7 Day average is higher than the 14 day average, it’s a sign that the case numbers have been rising over the last week.
This is the highest Spread Rate since Thanksgiving Week.
Looks like the reduction in restrictions + Covid Fatigue is taking its toll on Passaic. People are spreading more.
All formulas point to high 1400s. I’ll go with that.
Prediction: 1450
Prediction: 1300
Actual: 1337
Accuracy: 97.23%
Spread Rate for 3/1: 0.90
FORMULA FORECASTS
4 Week: 0.89 Spread Rate = 1194 Cases
3 Week: 0.94 Spread Rate = 1257 Cases
2 Week: 1.00 Spread Rate = 1337 Cases
14 Day Average: 202 Avg = 1414 Cases
7 Days Average: 191 Avg = 1337 Cases
MY PREDICTION
Looking at the 7 Day average again we see it’s lower than the 14 day. Hopefully this means the cases will continue a decline. Even with the slightly higher cases during 2/22, this past week’s cases are below 2/15s.
Looking at the decline over the last 4 weeks we can see a steady drop. This is why I am going to stick with my 4WASR forecast of 1194.
Prediction: 1150
Prediction: 2300
Actual: 2345
Accuracy: 98.08%
Spread Rate for 3/1: 0.90
FORMULA FORECASTS
4 Week: 0.94 Spread Rate = 2194 Cases
3 Week: 0.97 Spread Rate = 2265 Cases
2 Week: 1.03 Spread Rate = 2422 Cases
14 Day Average: 354 Avg = 2480 Cases
7 Days Average: 335 Avg = 2345 Cases
MY PREDICTION
All forecasts together get an average of 2341 which would be a 1.00 Spread Rate. But again, looking at the past 4 weeks we are seeing a decline. I’ll stick with the 4WASR forecast again.
Prediction: 2100
Not the best week. Last weeks accuracy was 74.47%
Total Project Accuracy now at 85.23%
Prediction: 2400
Actual: 1827
Accuracy: 76.13%
Spread Rate for 2/22: 6.74 (But it really doesn’t count since last week’s only recorded cases were 271 on Sunday)
Also, 2/28 only had 19 cases reported? Compared to the 7 day average of 261. So there’s a chance that data may change in the future like It has in Rhode Island a few times.
FORMULA FORECASTS
Formulas are basically useless this week, and probably next week, since there was barely any data recorded the week of the 15th thanks to the storm.
MY PREDICTION
Again, the storm really messed up all the data. Austin was on a good decline trend since 1/11.
2/8 only had 5 reporting days and recorded 1707 cases. (Would have been around 2300 cases if the two empty days recorded)
2/15 only had 1 reporting day and recorded 271 cases.
2/22 had 7 reporting days and recorded 1827 cases.
I expected 2/22 to be higher than 2/15 only my prediction was too high. But, if we look at the Rate of Decay from 2/8 (if it was 2300) to 2/22, we may be able to predict a decline of 250ish case next week.
Again, the data is all messed up so this one is more of a stab in the dark until a get a few consecutive weeks together again.
Prediction: 1600
Prediction: 950
Actual: 1201
Accuracy: 79.10%
Spread Rate for 2/22: 1.03
2/15 case count was 1167. Cases went up. Curious enough, the town has be holding sporting events last week and allowing parents to attend certain events.
FORMULA FORECASTS
4 Week: 0.92 Spread Rate = 1101 Cases
3 Week: 0.92 Spread Rate = 1104 Cases
2 Week: 0.88 Spread Rate = 1054 Cases
14 Day Average: 169 Avg = 1184 Cases
7 Days Average: 172 Avg = 1201 Cases
MY PREDICTION
Turns out the snow didn’t deter enough social interaction last week. Spread Rate went up by 0.03%
Didn’t I call this last week? I said the cases in Passaic like to plateau for two weeks then drop again. Did I just find a pattern? Only way to find out!
Last plateau was 2/1 & 2/8 when the cases stayed around 1600 and then dropped 440 cases. Before that the plateau was 1/18 & 1/25 when cases sat around 1800 then dropped 160.
Cases are either going to go up again as restrictions roll back and people socialize more, or they will drop like the last two plateaus.
Prediction: 1100
Prediction: 1100
Actual: 1490
Accuracy: 73.83%
Spread Rate for 2/22: 1.10
FORMULA FORECASTS
4 Week: 0.86 Spread Rate = 1288 Cases
3 Week: 0.89 Spread Rate = 1328 Cases
2 Week: 0.96 Spread Rate = 1433 Cases
14 Day Average: 203 Avg = 1421Cases
7 Days Average: 213 Avg = 1490 Cases
MY PREDICTION
Cases went up by 138. Not exactly what I expected. However I was curious last week if we might hit a Plateau in multiple cities as we hit a “Natural” case number. Meaning the virus was always going to get up to this number, but all the holidays caused a spike these last few months and we are now getting back down to “normal”.
Don’t ask why, but I think cases will go down slightly next week as the cold weather continues.
Prediction: 1300
Prediction: 1800
Actual: 2615
Accuracy: 68.83%
Spread Rate for 2/22: 1.17
2/15 was 2237, what are you guys doing over there???
FORMULA FORECASTS
4 Week: 0.90 Spread Rate = 2349 Cases
3 Week: 0.95 Spread Rate = 2480 Cases
2 Week: 1.00 Spread Rate = 2615 Cases
14 Day Average: 347 Avg = 2426 Cases
7 Days Average: 374 Avg = 2615 Cases
MY PREDICTION
I’ll be honest, I don’t know. Cases were at 2691 for 2/8, then 2237, then 2615. Based on this, I think the cases will stay in this range. The state is easing restrictions right now, so cases won’t drop but I doubt they will spike again. Unless you guys go crazy out there.
Prediction: 2300
There was a problem with the Rhode Island Data that was published last week. See the Rhode Island section below.
Actual accuracy rating for last week was 84.59
Due to the Texas situation and Rhode Islands data issue, I only counted the predictions for Passaic and Boston. Accuracy Rating is 89.76%
Due to the winterstorm that basically brought Texas to its knees (far reaching power black outs for five straight days, no drinking water, icy roads for four days, lack of food, no fast food, cats and dogs living together!!!!) there were no case updates from the 13th to the 20th. The first day back to regular reporting was yesterday the 21st.
Actual: 271
MY PREDICTION
Key Factors to look at:
1. The storm severely disrupted Covid reporting for the whole state.
2. People were stranded in their homes as the roads were almost impossible to traverse for a couple days.
3. People were without power for almost the whole week. Those with power opened their homes to people without.
4. There were “warming shelters” opened throughout the city for the homeless and those who didn’t have power or places to stay.
5. Once the power was restored, the temps rose again. I personally witnessed the crowds at bars and walking around town over the weekend as people celebrated the end of the freak storm.
So what does this all mean?
First of all, the numbers for the next week, maybe two weeks will be all off. My prediction formulas will basically be useless after missing a whole week of data and I’ll be flying blind. The power outage will result in three, somewhat contradicting, events:
1. People who should have gotten tested never will, therefore their cases will not be counted
2. There will be an overflow of older test results from before the storm, but never submitted due to the storm, that will hit this coming week’s numbers. This causes an artificial spike.
3. People who would have gotten tested during the storm will get tested this week. Basically two weeks’ worth of results will happen in one week.
Second, the increased socialization has the potential to cause a slight spike in the coming week and a half.
Example: People staying at “warming shelters” or with friends who had heat, power and food & then the celebratory lunches, dinners, drinks and hangouts over the weekend.
All of this together points towards a higher case count for the coming week or so.
Prediction: 2400
Prediction: 1400
Actual: 1167
Accuracy: 83.36%
Spread Rate for 2/15: 0.73
FORMULA FORECASTS
4 Week: 0.90 Spread Rate = 1049 Cases
3 Week: 0.88 Spread Rate = 1026 Cases
2 Week: 0.86 Spread Rate = 1009 Cases
14 Day Average: 198 Avg = 1387 Cases
7 Days Average: 167 Avg = 1167 Cases
MY PREDICTION
Currently there is snow on the ground in Passaic, NJ. There is a predicted mixture of snow and rain for tomorrow followed by three clear days and another weekend of potential storms. The weather, like last week, means reduced socialization and testing.
If this continues, Passaic may see its first sub-1000 case week since 10/26. But theres also the possibility that the cases plateau for a week like they did two weeks ago and result in a 1:1 spread.
Prediction: 950
SUFFOLK COUNTY (BOSTON, MASSACHUSSETTS)
Prediction: 1300
Actual: 1352
Accuracy: 96.15%
Spread Rate for 2/15: 0.82
FORMULA FORECASTS
4 Week: 0.79 Spread Rate = 1074 Cases
3 Week: 0.79 Spread Rate = 1062 Cases
2 Week: 0.79 Spread Rate = 1063 Cases
14 Day Average: 214 Avg = 1500 Cases
7 Days Average: 193 Avg = 1352 Cases
MY PREDICTION
Temps are low in Boston this week, but the weather won’t be too bad overall. Boston is on a good downward trend. Look at the below graph, the cases aren’t dropping as fast anymore, we are approaching what I consider to be a realistic case range. Meaning, if it weren’t for the Holidays and all those social events, the cases would probably have sat in this range naturally.
The end of the Holiday season resulted in an immediate and quick descent in cases. The Spread Rate also dropped. Look at this graph as well.
The Spread Rate seems to sit in the 0.80 area. I’m curious if this is the natural Spread Rate for Boston…. Is it possible to have a Natural Spread Rate???? I’m gonna research this idea more.
Prediction: 1100
When I updated the data for 2/15 I realized that the data for 2/8 was different than what I had recorded. There were an additional 1688 cases accounted for that I didn’t have. This isn’t the first time that Rhode Island has done this to me. They seem to have a problem with reporting their numbers in a timely manner. Not to mention they had some weather that week, so it’s possible there was a delay in reporting.
Their initial report for 2/8 was 1003 cases.
The cases for 2/1 was 3186. (this number did not change when the new update came in)
I predicted cases for 2/15, based on the reported 1003, to be 1200. Basically a Spread Rate of 1.20. I did this because I thought the 1003 was artificially low due to weather. Turns out I was right to doubt the number.
So I won’t count this week for my accuracy rating.
But I will update my accuracy rating for 2/8s prediction. I predicted 2400 cases and the updated actual was 2237, so my accuracy rating for 2/8 is actually 93.21%
Actual 2/15: 2237
Spread Rate for 2/15: 0.83 (When compared to the updated numbers for 2/8)
FORMULA FORECASTS
4 Week: 0.81 Spread Rate = 1811 Cases
3 Week: 0.81 Spread Rate = 1808 Cases
2 Week: 0.84 Spread Rate = 1875 Cases
14 Day Average: 352 Avg = 2464 Cases
7 Days Average: 320 Avg = 2237 Cases
MY PREDICTION
I’m weary thanks to the data mishap last week. Just looking at all the weekly formulas and the weather forecast, I’d say the 1800 area is probable.
Forecast: 1800
As the Covid numbers begin to drop across the country, I was curious what a timeline for actual reopening may look like. So I compiled a list of all the Covid Lockdown stages and various restriction orders that came out of Rhode Island. I then matched them up with the graphs I complied from all my data.
First we have the Positive Test Percentage graph. This graph shows the Percentage of Positive tests in all tests given. Please note that I have offset the percentages to the dates, meaning: The percentage at each Restriction update is the percentage of the week leading up to that date. This visualizes what the condition of the state was in that lead to the decision of each restriction update.
Next we have Total Hospitalizations. This is a pretty important graph as it shows the growth in which that amount of people are being admitted to the hospital for Covid. The Blue line is the Total number at that given date and the Red line is the growth number between restriction updates. Each date at the bottom is a date in which a Restriction update was released.
Finally we have the Estimated Current Hospitalizations graph. This one is a less reliable because the Hospitalization numbers are an estimate. I generated the graph by using only the new Hospitalizations for the last two-three weeks. I figured the current widely agreed recovery timeline for the virus is two-to-three weeks. That being said, this doesn’t count for any long term medical attention such as ventilators, so these numbers could be off.
Looking at all graphs we see a point in time where Covid seemed to have slowed down and then ramped back up. The graphs seem to point at the end of September / beginning of October, around the announcement of the state reducing gatherings from 25 to 15. “Covid Fatigue” and then the beginning of the Holiday season (starting with the notorious Halloween), set off a chain reaction that spiked cases and hospitalization for months. No further restriction could stop the rise in cases.
But that’s not to say those restrictions didn’t help keep the numbers where they were and not spiking even higher. The restrictions were like lightly using the brakes when you drive downhill: you aren’t going to come to a full stop right away, but you also aren’t going to gain more speed.
Furthermore, the controversial “Pause” was an interesting event. The “Pause” was enacted on November 30th, 2020 and ended on December 21st, 2020 and was designed to slow Covid Spread and to help flatten the curve of hospitalizations. But according to all the charts, it didn’t have a huge effect. Positive Test results went down barely half a percent. And then immediately shot back up past the point they were at before the “Pause”, then dropping continuously. Estimated Current Hospitalizations kept shooting upward, then the “Pause” ended at the first major drop. Which again, the hospitalizations jumped up again immediately and yo-yoed for a few weeks after. Even looking at Total Hospitalizations, the numbers jumped after the “Pause” was enacted and even seem to slow down their climb after it ends.
That being said, the Spread Rates for the weeks during the “Pause” were lower than previous weeks. But a Spread Rate is an indicator of 1:1 spread from each week. So when numbers were already at 8902 the week of 11/30, a Spread Rate of 0.96 the next week still resulted in over 8K new cases.
As the state reacted to small spikes and began to further restrictions, the cases kept climbing. The holidays were always going to be “Super Spreading” events, there is no argument there. But it appears that the restrictions didn’t help much in preventing them. Maybe just curbing them.
I for one believe Covid is real and a threat to those that are not healthy. That is why I look at the data. The data, I have found, paints a very clear picture of the virus and what is happening in each state and city. There are people who tell me that the sources I use (NYT, The Atlantic, Google) are all Left-Biased and that the sources report higher numbers than are real to help push a Liberal agenda. But if that were true, then the graphs would show me that the Rhode Island (a very Blue state) restrictions actual worked at stopping Covid. Instead it shows me the State struggling to keep up with the virus and human nature. It shows me that no one restriction was overly helpful in stopping anything. The only time that Covid slowed down was once all the social Holidays ended. In short, the virus stopped spreading because people didn’t have a reason to spread it.
Last week’s numbers hit way lower than I predicted. Except for Passaic. My drop in accuracy is mostly due to Rhode Island, they reported only 1000 cases which would mean there was only a 31% Spread Rate. I will keep an eye on the information coming out of RI in the next few days and report back.
Last Week’s Accuracy rating: 71.74%
Projects Overall Accuracy: 84.76%
Prediction: 2600
Actual: 1707
Accuracy: 65.65%
Spread Rate for 2/8: 0.55
There were no entries on the 13th or 14th. Had there been entries that matched the average daily case count then the weeks count would have been 2300.
FORMULA FORECASTS
4 Week: 0.78 Spread Rate = 1334 Cases
3 Week: 0.76 Spread Rate = 1294 Cases
2 Week: 0.69 Spread Rate = 1172 Cases
14 Day Average: 401 Avg = 2804 Cases
7 Days Average: 341 Avg = 2390 Cases
MY PREDICTION
All locations have artificially low numbers due to in climate weather. Austin is currently in the middle of a “Ice-Pocolapse.” People are without power all over the state and there is a potential for things to get worse. But next week (2/22) it will get warmer. Due to the current climate and closed businesses, I think the Covid count will be low for the next two weeks:
1. There will be a significant drop in virus spreading possibilities (Nowhere to shop, closed jobs, no outdoor socialization events)
2. Covid Testing facilities are closed due to the ice.
That being said, the rise in temps, delayed testing and potential for Covid spreading while people stormed the grocery stores this past weekend could result in Covid cases artificially spiking next week and potentially the end of this week.
Prediction: 1700 (again)
Prediction: 1380
Actual: 1606
Accuracy: 85.93%
Spread Rate for 2/8: 1.00
Passaic reported cases on all days.
FORMULA FORECASTS
4 Week: 0.95 Spread Rate = 1518 Cases
3 Week: 0.96 Spread Rate = 1535 Cases
2 Week: 0.96 Spread Rate = 1535 Cases
14 Day Average: 229 Avg = 1605 Cases
7 Days Average: 229 Avg = 1606 Cases
MY PREDICTION
Passaic County hit almost exactly where it was last week. The highest case reporting was on Wednesday, potentially test results coming back three days later from Monday. Monday would be spill over from the previous week. The rest of the week’s numbers fall in line with an area that’s slowly declining.
The area is supposed to receive more ice and snow this week.
Prediction: 1400
Prediction: 1760
Actual: 1647
Accuracy: 93.58%
Spread Rate for 2/8: 0.75
Boston reported cases on all days.
FORMULA FORECASTS
4 Week: 0.79 Spread Rate = 1303 Cases
3 Week: 0.79 Spread Rate = 1294 Cases
2 Week: 0.77 Spread Rate = 1264 Cases
14 Day Average: 274 Avg = 1919 Cases
7 Days Average: 235 Avg = 1647Cases
All weekly formulas are hitting in the same area because the last four weeks have seen a steady decline and no dramatic switches in any particular direction.
MY PREDICTION
As the temperature gets colder and the weather produces more snow and ice, the drop in Boston should continue. Currently it is supposed to snow on Thursday and Friday, which means reduced social interactions.
Prediction: 1300
Prediction: 2400
Actual: 1003
Accuracy: 41.79%
Spread Rate for 2/8: 0.31
Rhode Island did not report any cases on the 13th or 14th. Had they reported the daily average on those two days, the week count would have been around 1300.
FORMULA FORECASTS
4 Week: 0.62 Spread Rate = 623 Cases
3 Week: 0.63 Spread Rate = 628 Cases
2 Week: 0.53 Spread Rate = 533 Cases
14 Day Average: 299 Avg = 2095 Cases
7 Days Average: 143 Avg = 1003 Cases
MY PREDICTION
The cases this week dropped drastically. Mostly due to bad weather and cold temps. Would like to say that I think this big of a drop (69%) is rare. I will be looking into this more
As the temperature gets colder and the weather produces more snow and ice, the drop in Rhode Island should continue. Currently it is supposed to snow on Thursday and Friday, which means reduced social interactions.
But I think the numbers this week are artificial, theres only 1000 cases due to the weather. And while the weather will continue, I think we will start to realize actual case numbers this week.
Prediction: 1200
Here is a super quick run down of how Weather affects the Virus Spread Rate. The following graphs illustrate the Spread Rate Score ( Spread Rate goes up +1, Spread Rate does down -1) compared to the Temperature Score ( Temp goes up +1, Temp does down -1)
Example: A Spread Rate score of 1.52 is higher than 1.30 so the Spread Score gets a +1. But if the next week sees a Spread Rate of 1.05, thats lower than 1.52 so the Spread Rate gets a -1 and therefor cancels out.
There are many factors to consider:
Heat and/or Sunlight killing the virus
Higher Temps usually mean people go outside more
Lower Temps keep people inside
When it snows or gets really cold in New England, outdoor dining is nearly impossible
Austin climate is different from New England: Doesn’t get too cold, high temps could be in the 100s.
Austin is almost the polar opposite of New England and therefore the inhabitants of it react to weather differently. In the summer you can get temperatures of up to 105. In when temps go up, people stay inside with the Air Conditioning blasting. The outdoor socialization goes down. When temps start to go down, people spend more time gathering outside and preforming activities. While it seems to be reported that the virus doesn’t last long in sunlight, we can’t be sure that the increased socialization doesn’t create scenarios where the virus can spread affectively.
The graph clearly illustrates that, in Austin, the spread rate is directly affected by the temperature. Temps go up, Spread Rate goes down. Temps go down, Spread Rate goes up.
This graph is interesting. At first glance it doesn’t seem to be anything other than jumbled lines, but look again. There are clear indicators of the Temp and Spread Rates interacting. When Temps stayed high the Spread Rate rose, same when the Temps started to fall.
The lower temps on the East Coast bring snow and a complete lack of desire to go outside. So people stay inside, isolated from the outside world.
Temperature is the worlds greatest “Social Distancing” enforcer.
Boston has a mild case of causality. Temps rose but the Spread Rate didn’t continuously climb. That small spike of the Spread Rate to “1” is around the time they started to do outdoor dining in the city. But as temps started to drop it became increasingly harder and unpleasant to dine outside. Add the copious amount of snow that Boston just got and now you have people stuck inside, separated from people outside of their bubble.
Please Note: The Temp Score comes from the weather data station at TF Green and is what Providence uses for Daily Temp readings. I am not averaging the whole states temperature for this score.
Rhode Island is another mild case. But there is a pretty clear picture here: When Temps go down and people stay inside, cases go down.
As for the high temps in the summer and lack of Spread Rate Climb: Rhode Island was locked down pretty tight. That being said, the Spread Rate in Rhode Island was pretty high compared to other states, but the graph doesn’t show a continual climb.
Two weeks in a row, let’s go!
I had an accuracy rating of 91.07% last week. This plus last weeks 92.52% brings my Total Project Accuracy rating up to 86.16%
Prediction: 3375
Actual: 3100
Accuracy: 91.85%
Spread Rate for 2/1: 0.82
Based on this week’s performance, Austin won’t hit 100% population spread until 7/8/2028
FORMULA FORECASTS
4 Week: 0.96 Spread Rate = 2968 Cases
3 Week: 0.86 Spread Rate = 2661 Cases
2 Week: 0.86 Spread Rate = 2672 Cases
14 Day Average: 491 Avg = 3434 Cases
7 Days Average: 443 Avg = 3100 Cases
MY PREDICTION
Cases are headed down. Since Austin’s high of 4912 cases during the week of 1/11, cases are down 37%. That’s a big drop in only three weeks. I believe this trend will continue, but its going to get colder in Austin this coming week. As I’ve explained before, Temps go down and Cases go up. So this might prevent the cases from dropping to its next low number compared to nicer temps.
Looking at the Spread Rates since December, we can see a trend downward. The Trend line in this graph uses the averages of the current and previous week to illustrate a trend. The trend ends on 0.71 Spread Rate, which is roughly 2201 cases.
The last three weeks have seen new cases drop by, 14.89%, 9.89% and 18.26%. The average of those are 14.33%. If we see a 14.33% drop in cases this week, from last weeks 3100, then that will be 434 less cases or 2666 cases next week. Which both my 3WASR and 2WASR forecasts point to.
Prediction: 2600
Prediction: 1350
Actual: 1603
Accuracy: 84.22%
Spread Rate for 2/1: 0.91
Passaic was my worst performance last week. I guess the snow wasn’t enough to slow things down during the week, but it may affect future weeks. February 3rd only had 91 reported cases and the two days flanking it were lower numbers than the rest of the week too.
FORMULA FORECASTS
4 Week: 0.80 Spread Rate = 1280 Cases
3 Week: 0.93 Spread Rate = 1485 Cases
2 Week: 0.93 Spread Rate = 1496 Cases
14 Day Average: 240 Avg = 1683 Cases
7 Days Average: 229 Avg = 1603 Cases
MY PREDICTION
I’m looking at weather and the Spread Rates trending decline, also last week the cases dropped by 160. I’m inclined to say that this weeks cases will fall between the 4WASR and the 3WASR forecasts.
The average of those two comes out to roughly 1380. I’m hesitant but….i’m not saving lives here so whatevah.
Prediction: 1380
Prediction: 2100
Actual: 2191
Accuracy: 95.85%
Spread Rate for 2/1: 0.78
Boston was my best score last week! It also had the second lowest Spread Rate.
FORMULA FORECASTS
4 Week: 0.80 Spread Rate = 1761 Cases
3 Week: 0.80 Spread Rate = 1761 Cases
2 Week: 0.80 Spread Rate = 1759 Cases
14 Day Average: 356 Avg = 2495 Cases
7 Days Average: 313 Avg = 2191 Cases
Why are all the Weekly Spread Rate Forecasts so close to each other? The last four weeks have had very similar Spread Rates: 0.80, 0.81, 0.82, 0.78. This causes all the forecasts to be grouped together, since all the numbers they are pulling are similar.
MY PREDICTION
Snow, Snow come and stay. Put that Virus Spread Rate away…..
Last week’s Spread Rate was lower than the former three weeks. If this weeks Spread Rate stays at 0.78 then the new cases for this week would be 1716. I’d be willing to bet the case count lands in that area. But I’ll stick with my 4WASR.
Prediction: 1760
Prediction: 3450
Actual: 3186
Accuracy: 92.35%
Spread Rate for 2/1: 0.75
Rhode Island had the lowest Spread Rate last week!
FORMULA FORECASTS
4 Week: 0.84 Spread Rate = 2688 Cases
3 Week: 0.72 Spread Rate = 2304 Cases
2 Week: 0.78 Spread Rate = 2489 Cases
14 Day Average: 677 Avg = 4742 Cases
7 Days Average: 608 Avg = 4254 Cases
MY PREDICTION
The curfew for certain venues was lifted last week, followed immediately by snow fall. So if the curfew has an effect on the Spread Rate, we won’t see it as it gets counter acted. It snowed again yesterday in the area, so I believe the weather will continue to affect the Spread.
Cases have dropped by a 1000 for the last two weeks. So when I see a Forecast of 2304, I’m not surprised. I guess the question is…do we think this drop will continue at its current rate? It’s obvious that the weather is helping.
Also, looking at the last time Rhode Island was over 2500 cases in a week (4/20/20) the Spread Rate declined over 3 weeks and stayed in the 0.75-0.77 range for five weeks before summer plans started to push the numbers higher. Last week was the 3rd week of decline, so if April is any determination then Rhode Islands Spread Rate may rest in the 0.75 area for a little while. But in April the cases went up as temperatures went up, now temperatures are going down so the Spread Rate may not rest.
Fingers Crossed:
Prediction: 2400
I made up for that last time when I missed the mark. I had 92.52% accuracy last week, across all four locations. Yeeeeeeeeeeeeaaaaaa Booooooooooyyyyyyyy
I’m not bragging, you’re bragging….I’m just a Cow. Moo.
Moooving right along….
My total accuracy rating for this project is now: 85.49% Which is the highest it’s been since I started tracking four locations.
I’m no weather man, but I predict weather will play a theme in the coming weeks.
Prediction: 3600
Actual: 3768
Accuracy: 95.54%
Spread Rate for 1/25: 0.90
Looks like the virus is continuing its decline, but we can’t deny that the bad weather last week was a major player. No holidays to gather for, almost impossible to eat outside with the cold and then the Snow keeping people inside. The spread rate might have been 1.0 if not for those factors.
FORMULA FORECASTS
4 Week: 1.03 Spread Rate = 3877 Cases
3 Week: 1.00 Spread Rate = 3777 Cases
2 Week: 0.88 Spread Rate = 3302 Cases
14 Day Average: 567 Avg = 3974 Cases
7 Days Average: 538 Avg = 3768 Cases
All formulas agree the cases will be below 4K. The 3WASR formula is capturing 1/11s high 1.25 spread rate while the 2WASR is capturing the two lows of 1/18 & 1/25 (0.85 and 0.90).
The 14 Day average is on the high side because it picked up the tail end of the 1/11 spike that bled into 1/18s week = four days of cases hitting 700 and higher.
MY PREDICTION
As I found in Austin, Texas, the temperature plays a significant role in the virus spread rate from week to week. Temps go down, Spread Rate goes up and visa-versa. See below:
Spread Rate Score goes up or down by 1 based on if the Spread Rate was higher or lower than the week before.
Same goes for the Temp Score, if the Average temp was higher, the score goes up by 1 and visa-versa.
The nice weather (Average Temp is supposed to go up this week) and slowing Spread Rate should help keep cases down this week.
Just going by my gut instinct, I think the cases will land somewhere between the 3WASR and 2WASR forecasts. Actually, last week the cases dropped by 413….so maybe closer to the 2WASR.
Prediction: 3375
Prediction: 1600
Actual: 1763
Accuracy: 90.75%
Spread Rate for 1/25: 0.96
Passaic had the highest spread rate last week. That’s a good sign for turning this thing around.
FORMULA FORECASTS
4 Week: 1.42 Spread Rate = 2509 Cases
3 Week: 0.76 Spread Rate = 1343 Cases
2 Week: 0.94 Spread Rate = 1649 Cases
14 Day Average: 258 Avg = 1929 Cases
7 Days Average: 252 Avg = 1763 Cases
Most formulas predict lower than 2K cases. The 4WASR should be thrown out since its capturing 1/4s 3.41 spread rate. Which was the highest spread rate since March 30ths 3.47
MY PREDICTION
As I found in Austin, Texas, the temperature plays a significant role in the virus spread rate from week to week, see the graph above:
North Jersey just got absolutely dumped on with snow. This is going to stop people from going out, socializing, getting tested and probably slows down tests getting reported. So the numbers will be affected by this. Keep in mind, some of this week’s numbers will be pre-storm numbers being reported later. This storm will probably also effect next weeks (2/8) numbers, just not sure in which direction.
Due to the storm I think the 3WASR is probably close to the real number, maybe even lower???
Prediction: 1350
Prediction: 3000
Actual: 2798
Accuracy: 93.27%
Spread Rate for 1/25: 0.82
Boston had the second lowest Spread Rate last week.
FORMULA FORECASTS
4 Week: 0.95 Spread Rate = 2657 Cases
3 Week: 0.811 Spread Rate = 2268 Cases
2 Week: 0.814 Spread Rate = 2278 Cases
14 Day Average: 443 Avg = 3100 Cases
7 Days Average: 400 Avg = 2798 Cases
MY PREDICTION
Boston is different from Austin, Boston gets cold and stays cold for a while, Boston also gets more snow than Austin. Boston also knows how to drive in snow, which no one in Austin knows how to do and causes massive delays. Regardless…. We haven’t really seen the virus go up against a real New England storm, the kind where people don’t leave the house.
The virus seems to be slowing down from its holiday surge and the storms will add to that. Less socializing, less leaving the house and…..less testing.
Side note, over the last three weeks the cases have dropped by 1033, 819, 603 respectively. Following that trend and factoring in the snow, I’d imagine that number could be 700-800 next week.
Prediction: 2100
Prediction: 4700
Actual: 4254
Accuracy: 90.51%
Spread Rate for 1/25: 0.81
Congrats Rhode Island, you had the lowest Spread Rate last week!
FORMULA FORECASTS
4 Week: 0.91 Spread Rate = 3858 Cases
3 Week: 0.88 Spread Rate = 3724 Cases
2 Week: 0.71 Spread Rate = 3022 Cases
14 Day Average: 677 Avg = 4742 Cases
7 Days Average: 608 Avg = 4254 Cases
14 Day average is so high because 1/18s week had 4 days above 800 cases, 3 of which were over 900. Compared to last week which didn’t have a single day over 745.
Fun Note: Full Population Infection is projected for February 1st. 2025. Soooooo never.
MY PREDICTION
Rhode Island is going to be our most interesting study going forward. The state just lifted its curfew on restaurants and had many people celebrating by going out and socializing. No complaints here, however I am curious how that affects the spread rate. My initial reaction is to say the removal of the curfew caused celebration which exposes people to the risk of contracting Covid. Yet, the snow storm and cold weather may prevent people from actually going out. And it gets real cold in Rhode Island, I know, I shoveled there a few times.
So the lifted curfew and storm should clash. This makes me think the cases for next week will fall between the 3WASR and the 2WASR.
Prediction: 3400-3500 cases……Call it 3450