2023 年纽约州议会第 47 区特别选举结果

2023 年议会第 47 区特别选举共投出了 4442 票。 2447 票（55.1%）投给了民主党，1979 票（44.6%）投给了共和党。 0.3% 为空白。 民主党的胜率是10.5%，是过去10年来最小的胜率。

如果我们进一步观察登记选民，我们会发现，2022年11月，登记的民主党选民有35,442名，登记的共和党选民有10,501名，但到2023年，民主党登记选民为34,357名，共和党登记选民为10,386名。实质上，民主党失去了1,085名登记选民，共和党失去了115名登记选民 。 为什么？

三个原因：选区重划、疫情迁移、转党。

2022年，投票数为24,908票。 民主党获得 13,763 票（55.3%），共和党获得 9,937 票（39.9%）。 胜率是15.4%。 2020年，因为没有共和党候选人，所以得票率为0%，民主党得票率为69.3%。 所以，2020年有39.9%的选票是空白的。这些白票被称为“抗议票”。 2020年4.8%的选票是抗议票。 2023 年，白票进一步降至 0.3%。

这意味着共和党已经巩固了所有反现任选票。 从现在开始，选票占有率增加将选票从那些习惯性地：

1）投票给民主党的选民

2）很少投票的选民

后者的投票率要低得多，但共和党人更容易通过其立场的微小变化应合，而前者的投票率要高得多，但如果立场的微小变化也难以应合。

如果进一步考虑种族构成，该区40%的人口是白人: 1) 东北角希腊人、意大利人、斯拉夫人等白人较多，4): 南部则犹太人较多。 2) 和 3): 有华人和其他亚洲人，和西班牙裔。 未来的决定因素将是3) 的华人和其他亚洲人地区以及西班牙裔地区。

一旦竞选活动的财务数据和投票站的结果甚至选民档案被纳入，那么我们将更好地了解所发生的情况。

2023 Special Election Results of New York State Assembly District 47

4442 votes were cast in the 2023 Assembly District 47 Special Election.  2447 votes (55.1%) were cast for Democrats and 1979 (44.6%) were for Republicans.  0.3% were Blanks.  The margin of victory for Democrats is 10.5%, the smallest margin of victory in the last 10 years. 

If we look further at the registered voters, we find that there were 35,442 registered Democrat voters and 10,501 registered Republicans voters in November 2022, but 34,357 for Democrats and 10,386 for Republicans in 2023.  In essence, Democrats lost 1,085 registrations and Republicans lost 115 registrations.  Why?

Three reasons: redistricting, Pandemic moving out, Democrats to Republican conversion.

In 2022, 24908 votes were cast.  Democrats got 13763 (55.3%) and 9937 for Republicans (39.9%).  The margin of victory is 15.4%.  In 2020, there was 0 votes because there was NO Republican candidate and 69.3% for Democrats.  So, there were 39.9% of the votes were blanks in 2020.  These blank votes are called “protest votes”.  4.8% of the votes in 2020 were protest votes.  In 2023, Blanks are further down to 0.3%.  This implies that Republican has consolidated all anti-incumbent votes.  From now on, vote increase will have from voters who habitually:

1. Vote for Democrats
2. Rarely vote

The latter have a much lower voting rate but easier for Republicans to appeal with minor changes in its position while the former have a much higher voting rate but more difficult to appeal if only with minor changes in its positions.

If we further consider the racial composition, we will see that 40% of the population in the district are Whites, the northeast corner has more Greeks, Italians, Slavics and other Whites while the south has more Jews.  The north and north corner has Chinese and other Asians and the west has Hispanics.  The future determinant will then be in the Chinese and other Asians area and Hispanics area. 

Once the campaigns’ financial data and ED level results or even voter file are incorporated, then we will have an even better picture of what happened.

演讲稿：公共安全巡邏隊宣布擴大服務，纽约小房东合作，成资深顾问

2023年8月26日

11:30

4170 Main Street, 

Flushing, NY 11354

332字

李先生，你好！孟议员，你好！ 大家好！ 我是紐約小房東的會長何徳鄰。

今天是公共安全巡邏隊宣布擴大服務的重要日子。我想分享公共安全巡邏隊對我們不單是房東，而且是所有居民的重要性。請問大家還記得Tandika Wright嗎？她在今年三月二十七日因縱火殺人案被捕。此前，她有三十七次被捕記錄。而且，她更在二零二一年十二月攻擊亞裔而犯下種族仇恨罪。又或者周三發生的鐵錐殺人案，疑犯殺母傷子三人。

在如今罪案頻發，公權不彰，安全堪憂，獨善其身的社會，能有李先生以身作則，結隊營社，安邦衛民，貽範古今。 我們紐約小房東休戚與共，利害相連。有見公共安全巡逻队擴大服務，非但樂觀其成，更感桃李不言，下自成蹊，故毛遂自薦，野人奏曝。希望我們的貢獻積少成多，以不濟可。 再一次預祝公共安全巡邏隊安邦衛民，貽範古今。謝謝。

 

安全回家，票投安怡

2022年8月5日

2023年纽约州众议院第40选区共和党候选人廖安怡社区筹款餐会

Crown One, 34-20 Linden Pl, Queens, NY 11354

Sharon，你好！

侯会长，黄会长，陈会长，你好！大家好！

我是纽约小房东的会长，叫何徳邻。

今天是Sharon竞选的餐会。

我想问大家一个问题，其实好几个问题。

大家记得严志文是谁吗？记得吗，陈会长？记得Glenn Hirsch是谁吗？或者Melinda Katz?  这个记得了吧，Ron Kim?

严志文是一名外卖郎。他是今年4月28日送外卖的时候给Glenn Hirsch打死。为了什么？原来就是为了鸭酱。

严志文就跟我们一样，一个普通的人，他每天就是工作。想的，就是赚钱。他从来没说要小孩读什么哈佛耶鲁。很简单。就是赚钱，买房，跟小孩在一起。但是，就是工作的时候，回家的时候，给人打死。这样，你觉得他有什么错吗？没错吧？他无辜吗？绝对无辜。跟我们一样，就工作而已。

他也从来没想过什么堕胎控枪的。很简单的事情，对吧？你们有没有想过回家的时候，工作的时候回不了家？他就是这样的一个情况。

而且，那个疑犯，Glenn Hirsch，店家知道他的地址啊。但是，要等多久才把他拘捕？前前后后一共35天。疑犯你都知道在哪了，还可以等35天。才拘捕这个人。但是，拘捕了之后怎么样了？两天就放了他出来了。

是谁把他放出来了？

就是Melinda Katz。你知道Melinda Katz怎么样上台？2019年我们Queens有87万选民。投票给3万5，就是百分之4。百分之4的人投票给Melinda Katz。最后把疑犯放了的就是她。但这说明了什么？就是有百分之96的人没有投票（口误，百分之90的人没投票。其他5万5票由6个候选人瓜分）。百分之96的人的沈默让严志文的疑犯可以待35天，而且两天就给放出来了。所以，你不可以说你的那票没用。因为就是那个时候有百分之96的人没有投票。

现在让我们看看为什么那个时候Melinda Katz可以把他放出来？就是因为Ron Kim。

因为有一个保释改革法。Ron Kim投票的。Ron Kim让严志文的疑犯那么简单就给放了出来。Ron Kim今年为什么可以去大选？就是刚才侯会长说的，62票。他赢的就是62票才可以去大选。我们·今天是62票的好几倍吧？

如果我们想回家看小孩，麻烦你记得这一次一定要投票。因为，我们已经失去了第一个机会。现在，Sharon是我们最后的机会，最后的希望。如果大家以后工作的时候想回家的，记得要投票给Sharon。

谢谢。

（832字）

为何华社（包括教会在内）为自己生存需要推高投票率？

英稿中译

Why do Chinese institutions, including churches, have its own survival need to influence a higher voting participation rate?

More Chinese community members than in the last decades feel unsafe in NYC.  The fear of insecurity will increase its current trend of exodus, either to other states or to suburbs.  The consequence of this displacement will destroy the current institutions, including the Chinese churches, in Chinese communities. 

Insecurity, be it from a virus or from robbers or from mentally insanes, will make inter-personal relationships less personal in the name of protection or social distancing.  This is not simply moving church worships to the internet.  All church leaders know, Zoom worship means no worship.  The attendance via Zoom is in fact even lower than in person.  Internet convenience has not replaced in-person allegiance.  As people flee to Long Island or White Plains or FL, churches in Chinatown, Sunset Park and Flushing are dying. 

Many have shared the wisdom about inaction by leaders, moral leaders or otherwise.  Republic (Book 1, 347c) says “The price of political apathy is to be ruled by sinful men”.  “Wishing to maintain his personal purity allows that great morale to embroilment,” says Dialect, Weizi 7.  Proverb reminds us: When the righteous triumph, there is great glory, but when the wicked rise, people hide themselves (Proverbs 28:12). 

Community institutions bring not only service to its members, but also cohesion and identity.  Allowing disorder to permeate is encourage the flock to flee to an evil in the name of security.  Reminding members of an institution to vote is not only a moral or divine obligation, but also ensure the institution’s survival. 

过去几十年以来，华人在纽约不安全感达到最高点。治安败坏的恐惧将增加目前的人口外流趋势，无论是流向他州还是郊区。这种迁移的后果将摧毁目前在华社中的机构，包括华人教会。

不安全感，无论是来自病毒、仇恨犯罪还是精神病人，都会以保护或社交距离为名使人际关系变得不再人性化。这不仅仅是将教堂崇拜转移到互联网上。所有教牧都知道的：网礼拜就是没礼拜。实际上，网礼拜参与度低于人聚礼拜参与度。网便利并没有取代人情。随着人们逃往长岛、白原市或佛罗里达州，唐人街、日落公园和法拉盛的教堂正在消亡。

前人教导了我们不作为的智慧。柏拉图在共和国（Book 1, 347c）说“退出政治的代价是由罪人统治”。 “欲洁其身而乱大论”出自论语，微子 7。箴言提醒我们：義人得志，有大榮耀；惡人興起，人就躲藏（箴言 28:12）。

社区机构不仅为其成员带来服务，还带来社会凝聚力和身份认同感。允许治安败坏是鼓励信徒逃往魔鬼以安全之名所建的邪恶。提醒会众投票不仅是一种道德或神的使命，而是确保教会的生存。

Protected: 美國新症何時到最高日？

This content is password protected. To view it please enter your password below:

Password:

Is New York’s “Stay Home Order” issued on 3/20 Effective?

New York state has been having the most infected for days and therefore is the greatest contributor to the US’s culminated infected.  We will focus on the recent development of NY’s newly infected.  After reaching 5707 cases on 3/23 Monday, NY has been reporting fewer cases than 3/23 on following consecutive days, 4790 on 3/24 and 5146 on Wednesday.

China imposed its national quarantine on 1/20 and witnessed its peak on 1/27, 7 days afterwards.  NY’s Stay Home Order is much weaker than that.  And it is now taking into effect in 5 days, much shorter than the more restrictive quarantine.

Is “Stay Home Order” issued on 3/20 is taking into effect in NY?

We are hopeful that it is working, if the following question is resolved.

NY has been reducing tests since 3/23.  In fact, if one wants to have zero newly infected cases, then one can simply perform zero test to achieve that result.  How does having fewer tests show that the peak is over?  Figure 1 shows the tests performed and newly infected since 3/17.  As of 3/25’s data, NY has tested 103,479 patients.  Among those tested, 30811 are positive, 29.8%.  Put it in perspective: Korea tested 349,000 patients, and 9,037 positives, i.e. 2.59%.

Figure 1: Tests Per Day (Blue) and Newly Infected Per Day (Orange) Since 3/17

Korea has successfully controlled this epidemic without having a quarantine because they identified and subsequently treated the infected.  For every one positive patient in Korea, they checked another 37 people.  For every positive patient in NY, only 2.4 other patients are checked.

Tests seem to have stablized at 12,000 per day.  However, the rate of infected among tested is actually increasing.  As of 3/25, the positive among tested on that day is now 42% from 34% on 3/23 when the peak was recorded and 29% from 3/20 when Stay Home Order was issued.

Figure 2: Newly Infected (Blue) and Newly Infected Rate Among Tested Patients (Orange) Since 3/17

It is true that NY has fewer positives identified.  How many potential patients are left unchecked?

 

Data:

1. https://docs.google.com/spreadsheets/u/2/d/e/2PACX-1vRwAqp96T9sYYq2-i7Tj0pvTf6XVHjDSMIKBdZHXiCGGdNC0ypEU9NbngS8mxea55JuCFuua1MUeOj5/pubhtml#
2. https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_the_United_States
3. https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_South_Korea

When Will We See Coronavirus Reaching Its Turning Point?

We suddenly reached 2000 culminated infected of coronavirus in the United States from merely 50 in a matter of 2 weeks.  In fact, the newly infected has been increasing every day.  On March 3^rd, it was 85.  The following day grew 26 cases.  From March 4^th’s 111, it went up by 56 cases on March 5^th to 175.  Between March 11 and March 12, 395 new cases are reported.

Figure 1: Culminated and Newly Confirmed Infected in Table

Figure 2: Culminated and Newly Confirmed Infected in Graph

March 1^st is the day when China has only half of the patients still in treatment.  It now has less than 15% of the patients remain hospitalized.  The end is within reach in China.  We don’t even know when our peak will come, let alone the end.  How will we know if the peak is approaching?

Now foreign experience, namely China and Korea, may come into handy.

Even when China imposed as strong an intervention as a quarantine not only an entire city (Wuhan has 11 million, NYC has 8), but the entire country on January 23, the culminated infected was still climbing for another 10 days due to the 14 day incubation period of the virus.  One may want to note that the peak on 2/12 is actually due to new testing method.  The 2/4 peak is of our concern and not the 2/12 peak in this study.

Figure 3: Culminated and Growth Infected in China

However, we are not getting a quarantine in the States yet.  But if we will, Chinese data show us 14 days and the highest growth will pass.

Short of having a quarantine, what else can we learn from Chinese and Korean data?

Yes, there are other things we can observe.  If we compare the graphs of the US and Chinese culminated infected and growth, we would realize that the growth in the US has not decreased yet.  In fact, the growth has been increasing.  However, Korea shows they have slowed down their growths.  So, we want to observe if there is a metric that can at least pre-date if not predicted the behavior of the growth of infected in USA.   And the culminated infected of these two countries show a consistent behavior we may want to recognize.  The below two graphs show their newly infected per day follow the pattern called probability density function of logistics distribution.

Figure 4: Newly Confirmed Infected in Korea and Probability Density Function of Logistics Distribution

Figure 5: Newly Confirmed Infected in China and Probability Density Function of Logistics Distribution

 

The peak means after that point, the newly infected per day (growth) will decrease consistently.  In other words, that is a turning point.  So, our job is to find when this turning point will happen.  However, studying the growth will introduce a bias toward the large numbers: the growth of 100 when yesterday’s culminated infected is 200 (100/200 = 50%) versus 10,000 (100/10,000 = 1%) makes a big difference.  Therefore, in order to have a more equal weight of these growths, we look at the growth rate instead.  Here is the growth rate from Chinese data, prior to Chinese new testing implementation reported on 2/10.  Notice the peak occurs when the growth rate of infected is approximately 25% (right side vertical axis)?

Figure 6: China Growth and Growth Rate

We check it again on the Korean data.  And we approximated the growth rate as a function of negative exponential function, shown as the grey line in the below graph, in order to verify how close it would be between the time the growth rate is expected to be 25% and the time of peak.  It turns out the peak occurs on 2/29, the approximated growth rate of infected , via exponential decay, occurs on 2/28, and the actual growth rate of infected on 2/29 is 30%.

Figure 7: Korea Growth and Growth Rate

Armed with this information, one can now estimate when the peak of growth occurs in the United States.  One would hope to have a second method to provide one more perspective.  Second difference moving average would be an ideal method.  However, since growth rate has been increasing and has not been stable, some predictions by this method tell us that we will never reach the peak.  This method is sensitive to data.  Therefore, it only gives informative results when data are stable.  Below is the chart (Figure 8) that each method tells us how many days away we are from the peak, depending the latest data one would use.

Figure 8: Predicted Peak Dates by Method and by Data As Of Date

If one makes predictions on March 7 with the exponential decay method on the growth rate, then one may predict the peak cannot arrive before 3/11.  Of course, it certainly works but we would like to see the prediction date closer to the actual.  So, multiple predictions are made.  Similarly, if one uses the 2nd difference moving average method with data up to March 11, then the peak is not expected to arrive before 3/12.  The peak dates by the prediction date and method are tabulated in Figure 8. 

One may also notice that the 2^nd difference moving average gives us predictions closer to the prediction date than the exponential decay method.  This  characteristic is common between these two methods.  However, what is of interest to note is that the more stable method, exponential decay on growth rate, although gives more informative results, the results are telling us the peak growth is getting farther and farther away from as us as we get more data.  On March 6, the turning point is expected to be 3 days away.  By the time when prediction is made on March 13, the turning point is then 5 days away.  When a method is known to give more stable results turns out to be giving diverging results, it simply means we are losing control of this situation, if we ever were.

Based on the two different methods, we can be certain the peak of growth will not take place before those prediction dates.

 

 

Data:

USA: https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_the_United_States

Korea: https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_South_Korea

China: https://news.ifeng.com/c/special/7tPlDSzDgVk

What is the Culminated Infected in USA on 3/12?

Globalization has brought us advantages as wells disadvantages, although we have seldom considered the disadvantages other than higher liquidity of jobs (i.e. moving job opportunities to the other side of the globe).  Epidemic is now realized as another disadvantage as global travel becomes more common.  Not long ago, Coronavirus was still a distant harm.  It is now a clear and present danger to anyone regardless of geography.

In the earlier articles, we predicted the fatality rate among the concluded patients as well as the proportion of patients remain in treatment.  Can we predict the culminated infected in the near future?  How far ahead can we predicted?

The earlier articles used two methods for prediction:

• Second order difference moving average
• Exponential function

In this article, we realize that second order difference moving average method is sensitive to input data.  In this case, the data are too unstable to enable this method to give us good results. From the below graph (Figure 1), one can easily conclude that exponential function is a good choice for prediction.  Therefore, we will focus on the predictions by exponential function.

Most prefer to have as many data records as possible.  The general idea is that if we have more data, we will generate better prediction.  We will check if this idea works to our favor in this article.

However, having too much data also has a trade-off: the model become insensitive to the more recent data.  Therefore, in this evolving coronavirus scenario, it becomes even more important that we have to identify the optimal days of records we want to use to generate the statistical results based on how many days ahead we want to predict.

We use three days of data to predict the next few days’ culminated infected in the States.  We then use four days for predictions, five days, up to fourteen days to see the error rates.  Therefore, if we want to make predictions for March 7^th, we collect the data from March 4^th to March 6^th (3 days) to make prediction for March 7^th, also March 3^th to March 6^th (4 days) to make prediction for March 7^th, March 2^nd to March 6^th (5 days) to make prediction for March 7^th, etc.  Thus, there are 12 experiments where the first prediction date is March 7^th.  Just in case March 7^th happens to be more favorable to the data set, we also try another 12 experiments where the first prediction date is March 6^th, ranging from 3 days in a sample to 14 days in a sample.  We try six different first prediction dates.  In other words, we have 6 6 experiments of 3 days sample, 6 experiments of 4 days sample, etc.  Therefore, we have a total of 72 experiments to test what makes better.

These experiments then make predictions to new data which are not part of the data samples.  Errors are then measured to examine their accuracies.  The last day of data used for these experiments is March 6^th.  The last day of data to be used to make predictions for March 12^th is March 11^th.

The way to measure if one is better than another is by comparing the difference between the maximum and minimum prediction’s error percentages (MaxMin) within the sample against another.

In the below Figure 2, we observe four different lines where each one measures the MaxMin.  A MaxMin line connects the averages of the MaxMin among those 6 experiments having the same quantity of days in the sample across the varying from 3 days to 14 days in the sample.

So, we find that for Day 1 predictions, having only the most recent data (namely 3 and 4) does not give us better predictions since those 3 day samples give us a MaxMin that above 20%.  Five days samples give us better results.  When we increase the sample from 6, we get high MaxMin and eventually peak at 10.  MaxMin then decreases as we get more data records in the sample.

In fact, samples of 5 days of data give us lower MaxMin for Day 2 predictions, Day 3 and Day 4 too.  However, having 5 days of data maybe too aggressive.  Therefore, we opt to take 11 days and 12 days samples to make our predictions.

They give us MaxMin much lower than the peaks, the data are not as long as 14 days and not as aggressive as 5 days.

Exponential function typically over predicts because the more recent data greater in magnitude and minimizing the errors means bigger numbers in the data set have a larger influence, be it exponential increase or exponential decay.  Therefore, we use the predictions as upper bounds.  We then use the average MaxMin as the lower bounds.  As culminated infected increases, the error produced by exponential function increases at a greater speed.  Therefore, we are interested at looking for a lower bound as well.

The predictions from 3 days sample to 14 days sample are presented in Figure 3.

The lowest upper bound is produced 983 by the 14 day sample.  The largest upper bound is 2451 by the 4 day sample.  The lowest lower bound is 639 by the 14 day sample whereas the largest lower bound is 2069 by the 4 day sample.  The 11 day and 12 day samples produce 1011 and 1273 lower bounds.

Human intervention will affect the culminated infected predictions, such as quarantine.  Quarantine may be in different forms other than the kind practiced in China where people are mandated to remain at home to be enforced by police.  NY has called on National Guards at New Rochelle.  Companies have mandated employees to work from home.  Some schools are closed in New York which effectively make parents stay at home as well.  These are not quarantine per se but are definitely human intervention which affect the culminated infected.

 

Fatality Rate on 3/1 in China

First published in Chinese on 2020/02/17.  The original can be found here.

Over the past four weeks, the number of deaths has increased dramatically. The fatality rate from discharge was as high as 60%. Now it drops to 15%. What will be the final fatality rate at 3/1?

 

Regarding the definition of fatality rate, please consider the definition on Wikipedia: https://en.wikipedia.org/wiki/Case_fatality_rate

 

The previous article has analyzed that because the epidemic is still developing and does not meet the conditions for calculating the correct fatality rate, we propose a method for calculating the discharged fatality rate before all cases reached their conclusions and thus fit for the correct fatality rate formula:

 

Case fatality rate = ;

 

To predict future discharged fatality rates, we chose two methods of analysis:

1. Exponential decay
2. Second-order difference 3-day moving average

 

The first is the processing of data. We exclude the data before (1/27) beforehand.

Exponential decay model requires the data to be monotonically decreasing. In other words, the first-order difference in exponential decay is required to be negative. Although Wuhan quarantine starts on 1/23, its impact is only observed starting on 1/27. Four days after the replacement of the closure policy, the spread of the virus is damped. The quarantine is effective. The data prior to quarantine is monotonically increasing, so the previous data is gradually replaced by the exponential decay method. Therefore, we excluded data before 1/27. The “second-order difference 3-day moving average method” only requires three days of data for each computation, so deleting the data before 1/27 does not affect this method. In this way, we use two methods to obtain the discharged fatality chart of Figure 1.

 

Figure 1 Discharge fatality rate curve (1/23 Wuhan quarantine, 2/9 the first batch of medical teams arrived in Hubei)

 

In Figure 1, the exponential decay (2/14) curve is obtained by the exponential decay method, the end date of the data used is 2/13. Similarly, the second-order difference 3-day moving average (2/16) curve is obtained by using the “second-order difference 3-day moving average” method. The end date of the selected data is 2/15, which is 2/13, 2/14, 2/15 data. In this way, the second-order difference 3-day moving average (2/17) curve is different from the second-order difference 3-day moving average (2/16) curve, except that the selected data is three-day data up to 2/16 (2/14, 2/15, 2/16 three-day data).

 

From Figure 1, we can observe that although the data differ by one day only, the second-order difference 3-day moving average (2/16) curve and the second-order difference 3-day moving average (2/17) curve are very different. Therefore, the second-order difference The 3-day moving average method is particularly sensitive to changes in data.

 

In order to study the effect of data changes on the predicted value curves of the second-order differential 3-day moving average method, we predict with four sets of data to obtain the discharged fatality rate graph of Figure 2. The resulting discharged fatality rates from March 1, 2020 vary widely. Among them, the latest data from 2/14 is used to predict that the discharged fatality rate will reach 11% on March 1, 2020, and that the SARS fatality rate is about 11%. The 2/16 curve reached a negative value.

 

Fig. 2 Discharged fatality rate by the second difference moving average

 

Why is the 2/16 prediction so abnormal and negative? We analyzed the data carefully. Table 1 was obtained.

Table 1 Discharged fatality rate predicted by the second-difference 3-day moving average

 

As can be seen from Table 1, the hospital fatality rate decreases from 18.78% of 2/12 to 17.55% of 2/13. The first-order difference is changed from -0.23% to -1.24%, the second-order difference is changed from 1.02% to -1.00, and the three-day shift is converted from at least 0.01% (= (1.02% +-1.00% + 0.01%) / 3) 0.20% (= (1.00% + 0.01% + 0.38%) / 3). On 2/14, the three-day move may still be affected by 1.02% (the first item in the scarlet letter) of 2/12. Therefore, the 2-day difference of 2/14 has a small three-day shift of only 0.01%. However, on 2/15, 1.02% of the effect is removed from the calculation, and the difference from 2/14 and 2/15 is very small, which makes the penetration change of 2/13 become the second order three days of 2/15 Predicted change in moving average at 2/16 (black box). In the next few days, the second-order difference becomes smaller again, which is an error in the convergence of the second-order difference. Since the 2-day difference three-day moving average requires only 3 days of data, the prediction of 2/16 is abnormal and only related to the data of 2/13, 2/14, and 2/15.

 

The unusual data point of 2/13 causes the prediction of 2/16 to be unusual. Why is the data point on 2/13 unusual? It probably is a system disturbance caused by exogenous information.  The largest exogenous factor is the arrival of medical teams from other provinces at Wuhan, which is a quarantined city at that time. The first batch of medical teams arrives Wuhan on 2/9.  Perhaps due to a backlog of documents, it may take 4 days to reflect to the data, the unusual data behavior on the 2/13. Coincidentally, it takes 4 days for the quarantine’s effect to be observed in the data.

 

Using the second-order difference 3-day moving average method to be sensitive to data, we can obtain a new forecast curve based on new daily released data. We can then obtain an interval of predicted values for the discharged fatality rate of 3/1.

 

Using the exponential decay method, the resulting hospital fatality rate curve is shown in Figure 3. No matter how the data change, the exponential decay method produces similar curves. Therefore, the exponential decay method is not sensitive to the data, and the other data do not change the direction of the curve, so you can use the exponential decay as a relatively stable prediction. According to the prediction of this method, a fatality rate of 3/1 can be obtained below 11%.

 

Figure 3 Discharge fatality rate curve obtained by exponential decay method

 

We choose two methods, one is sensitive to data changes, and the other is more stable. In the end, we got a set of 2020/3/1 discharged fatality predictions, as shown in Table 2.

Table 2 Predicted 3/1 discharged fatality from different methods

 

We can observe that the exponential decay also produces a very stable prediction from 11.64%, and the emergence of new data for conversion continues to decline by 0.30% every day. New data should continue to push the new predicted values to decline. Moreover, the 3-day moving average method should be relatively sensitive to new data. The range of predicted values ​​vary widely, and aside from the abnormal predictions of 2/16, we can expect the lower limit of hospital fatality to be 3.05%.

 

According to our forecasts, by 2020/3/1, the discharged fatality rate range is between 9% and 4%.

---

The actual discharged fatality rate on 3/1 is 6.15% (=2915/(2915+44518)).  The mid point between 4% and 9% is 6.5%.