We would like to present the English version of the math model of AFU losses.
Past week we posted an article in Russian and presented our estimations for Armed Forces of Ukraine (AFU) losses and total personnel numbers. Now that we have received feedback, we present our translation of the entire article.
In it we apply statistical methods to our database of more than 125,000 records to produce independent estimations of how much people are fighting on the side of Ukraine, how much were killed and how much were captured. The model is based solely on the data we have on our hands and does not rely on any third-party guesses or estimations.
That being said, our estimations on 03.11.2022:
- killed in action: 106,209 combatants (+5,686 since the past week)
- captured: 24,348 combatants (+1,408 since the past week)
- active AFU personnel: 364,919 combatants (+8,497 since the past week).
The details and charts with daily numbers are available in the publication on our website.
#mathmodel_en
Past week we posted an article in Russian and presented our estimations for Armed Forces of Ukraine (AFU) losses and total personnel numbers. Now that we have received feedback, we present our translation of the entire article.
In it we apply statistical methods to our database of more than 125,000 records to produce independent estimations of how much people are fighting on the side of Ukraine, how much were killed and how much were captured. The model is based solely on the data we have on our hands and does not rely on any third-party guesses or estimations.
That being said, our estimations on 03.11.2022:
- killed in action: 106,209 combatants (+5,686 since the past week)
- captured: 24,348 combatants (+1,408 since the past week)
- active AFU personnel: 364,919 combatants (+8,497 since the past week).
The details and charts with daily numbers are available in the publication on our website.
#mathmodel_en
[Русская версия]
Our estimations on 10.11.2022 based on the model we published:
- killed in action: 109,824 combatants (+3,615 since the past week);
- captured: 24,390 combatants (+42 since the past week);
- active AFU personnel: 342,803 combatants (-22,116 since the past week: we have actively been cleaning the duplicate records which affected the estimation gravely. Also we would like to remind that the growth of KIA can be caused by the records added on past losses which means not all those people died this week).
Project statistics:
- the number of the confirmed records: KIA - 4,270; captured - 5,504;
- search requests received: 3,891 (multiple requests on the same person are counted as a single request);
- for 1,024 requests we provided some information,
- probability to find something upon a request: 26.3% (roughly 1 in 4).
#statistics #mathmodel
Our estimations on 10.11.2022 based on the model we published:
- killed in action: 109,824 combatants (+3,615 since the past week);
- captured: 24,390 combatants (+42 since the past week);
- active AFU personnel: 342,803 combatants (-22,116 since the past week: we have actively been cleaning the duplicate records which affected the estimation gravely. Also we would like to remind that the growth of KIA can be caused by the records added on past losses which means not all those people died this week).
Project statistics:
- the number of the confirmed records: KIA - 4,270; captured - 5,504;
- search requests received: 3,891 (multiple requests on the same person are counted as a single request);
- for 1,024 requests we provided some information,
- probability to find something upon a request: 26.3% (roughly 1 in 4).
#statistics #mathmodel
Telegram
wartears
[English version]
Оценка на основе математической модели WarTears.org на 10.11.2022:
- потери ВФУ погибшими: 109 824 человек (+3 615 за неделю);
- пленными: 24 390 человек (+42 за неделю);
- активный состав ВФУ: 342 803 человек (-22 116 за неделю: на этой…
Оценка на основе математической модели WarTears.org на 10.11.2022:
- потери ВФУ погибшими: 109 824 человек (+3 615 за неделю);
- пленными: 24 390 человек (+42 за неделю);
- активный состав ВФУ: 342 803 человек (-22 116 за неделю: на этой…
[Русская версия]
Our estimations on 10.11.2022 based on the model we published:
- killed in action: 113,380 combatants (+3,556 since the past week);
- captured: 24,681 combatants (+291 since the past week);
- active AFU personnel: 338,845 combatants (-3,958 since the past week).
Project statistics:
- the number of confirmed records: KIA - 4,374 (+104); captured - 5,600 (+96);
- search requests received: 4,018 (+127) (multiple requests on the same person are counted as a single request);
- for 1,068 (+44) requests we provided information,
- probability to find something upon request: 26.6% (+0.3%) (roughly 1 in 4).
#statistics #mathmodel
Our estimations on 10.11.2022 based on the model we published:
- killed in action: 113,380 combatants (+3,556 since the past week);
- captured: 24,681 combatants (+291 since the past week);
- active AFU personnel: 338,845 combatants (-3,958 since the past week).
Project statistics:
- the number of confirmed records: KIA - 4,374 (+104); captured - 5,600 (+96);
- search requests received: 4,018 (+127) (multiple requests on the same person are counted as a single request);
- for 1,068 (+44) requests we provided information,
- probability to find something upon request: 26.6% (+0.3%) (roughly 1 in 4).
#statistics #mathmodel
Telegram
wartears
[English version]
Оценка на основе математической модели WarTears.org на 17.11.2022:
- потери ВФУ погибшими: 113 380 человек (+3 556 за неделю);
- пленными: 24 681 человек (+291 за неделю);
- активный состав ВФУ: 338 845 человек (-3958 за неделю).
Статистика…
Оценка на основе математической модели WarTears.org на 17.11.2022:
- потери ВФУ погибшими: 113 380 человек (+3 556 за неделю);
- пленными: 24 681 человек (+291 за неделю);
- активный состав ВФУ: 338 845 человек (-3958 за неделю).
Статистика…
[Русская версия]
Our estimations on 24.11.2022 based on the model we published:
- killed in action: 123,215 combatants (+9,835 since the past week);
- captured: 25,704 combatants (+1,023 since the past week);
- active AFU personnel: 335,654 combatants (-3,191 since the past week).
Project statistics:
- the number of confirmed records: KIA - 4,749 (+375); captured - 5,775 (+175);
- search requests received: 4,173 (+155) (multiple requests on the same person are counted as a single request);
- for 1,103 (+35) requests we provided information,
- probability to find something upon request: 26.4% (-0.2%) (roughly 1 in 4).
#statistics #mathmodel
Our estimations on 24.11.2022 based on the model we published:
- killed in action: 123,215 combatants (+9,835 since the past week);
- captured: 25,704 combatants (+1,023 since the past week);
- active AFU personnel: 335,654 combatants (-3,191 since the past week).
Project statistics:
- the number of confirmed records: KIA - 4,749 (+375); captured - 5,775 (+175);
- search requests received: 4,173 (+155) (multiple requests on the same person are counted as a single request);
- for 1,103 (+35) requests we provided information,
- probability to find something upon request: 26.4% (-0.2%) (roughly 1 in 4).
#statistics #mathmodel
[Русская версия]
Those who follow our math model for AFU losses have most likely noted the sharp drop in the estimated number of killed in action in the past three days. Two days ago we noticed a small decline. When we saw the drop we decided to look deeper into what's going on.
What happened: we started to process official lists of soldiers awarded posthumously by president of Ukraine. The lists contain a few thousand names, but we decided to match them with the requests for search we have from relatives first.
We matched some succesfully. 26 matches found just yesterday are 18% of the whole number of dead prevously found on request by our project. We need to state here that total numbers of found dead and found captured are 166 and 996 (63 exchanged), respecrively. In our math model this number goes into denominator, hence our estimation dropped by exactly this 18%. Along with the other added data on the day it reduced the estimation by ~16k.
But then we talked to relatives, peered more carefully into the records, and found a whole bunch of alarming stories. After a discussion we decided not to roll back the changes, but to pause further data intake from this very source.
We hope to figure out how much we can trust it soon. In meantime we'd like to share one of the stories.
A soldier served in a mechanised brigade went missing. He was awarded posthumously on 8 August by the presidential decree №323/2022. Only 10 days after did the relatives get an official death notice. Some more time later the body of the deceased were returned to them. It turned out the body was of a different person.
#statistics #mathmodel
Those who follow our math model for AFU losses have most likely noted the sharp drop in the estimated number of killed in action in the past three days. Two days ago we noticed a small decline. When we saw the drop we decided to look deeper into what's going on.
What happened: we started to process official lists of soldiers awarded posthumously by president of Ukraine. The lists contain a few thousand names, but we decided to match them with the requests for search we have from relatives first.
We matched some succesfully. 26 matches found just yesterday are 18% of the whole number of dead prevously found on request by our project. We need to state here that total numbers of found dead and found captured are 166 and 996 (63 exchanged), respecrively. In our math model this number goes into denominator, hence our estimation dropped by exactly this 18%. Along with the other added data on the day it reduced the estimation by ~16k.
But then we talked to relatives, peered more carefully into the records, and found a whole bunch of alarming stories. After a discussion we decided not to roll back the changes, but to pause further data intake from this very source.
We hope to figure out how much we can trust it soon. In meantime we'd like to share one of the stories.
A soldier served in a mechanised brigade went missing. He was awarded posthumously on 8 August by the presidential decree №323/2022. Only 10 days after did the relatives get an official death notice. Some more time later the body of the deceased were returned to them. It turned out the body was of a different person.
#statistics #mathmodel
[Русская версия]
Our estimations on 08.11.2022 based on the model we published:
- killed in action: 112,919 combatants (*);
- captured: 26,900 combatants (+187 since the past week);
- active AFU personnel: 339,716 combatants (*).
Project statistics:
- the number of confirmed records: KIA - 5,019 (+270 in the past two weeks); captured - 5,772 (-2 in two weeks, a few #exchange happened);
- search requests received: 4,268 (+95 in two weeks) (multiple requests on the same person are counted as a single request);
- for 1,220 (+117) requests we provided information,
- probability to find something upon request: 28.6% (+2.2%) (roughly 1 in 4).
(*) - we choose not to publish weekly change as they are mostly caused by pseudo-dead from lists of posthumously awarded (we're still working on them). Another factor in the slight lowering of our estimations is the noticeable growth of us successfully finding people whose relatives requested to search for.
#statistics #mathmodel
Our estimations on 08.11.2022 based on the model we published:
- killed in action: 112,919 combatants (*);
- captured: 26,900 combatants (+187 since the past week);
- active AFU personnel: 339,716 combatants (*).
Project statistics:
- the number of confirmed records: KIA - 5,019 (+270 in the past two weeks); captured - 5,772 (-2 in two weeks, a few #exchange happened);
- search requests received: 4,268 (+95 in two weeks) (multiple requests on the same person are counted as a single request);
- for 1,220 (+117) requests we provided information,
- probability to find something upon request: 28.6% (+2.2%) (roughly 1 in 4).
(*) - we choose not to publish weekly change as they are mostly caused by pseudo-dead from lists of posthumously awarded (we're still working on them). Another factor in the slight lowering of our estimations is the noticeable growth of us successfully finding people whose relatives requested to search for.
#statistics #mathmodel
[Русская версия]
Our estimations on 15.12.2022 based on the model we published:
- killed in action: 121,954 combatants (+9,035 since the past week);
- captured: 27,804 combatants (+904);
- active AFU personnel: 337,649 combatants (-2,067).
Project statistics:
- the number of confirmed records: KIA - 5,441 (+422); captured - 5,894 (+122);
- search requests received: 4,648 (+380) (multiple requests on the same person are counted as a single request);
- for 1,254 (+34) requests we provided information,
- probability to find something upon request: 27.0% (-1.6%, roughly 1 in 4).
#statistics #mathmodel
Our estimations on 15.12.2022 based on the model we published:
- killed in action: 121,954 combatants (+9,035 since the past week);
- captured: 27,804 combatants (+904);
- active AFU personnel: 337,649 combatants (-2,067).
Project statistics:
- the number of confirmed records: KIA - 5,441 (+422); captured - 5,894 (+122);
- search requests received: 4,648 (+380) (multiple requests on the same person are counted as a single request);
- for 1,254 (+34) requests we provided information,
- probability to find something upon request: 27.0% (-1.6%, roughly 1 in 4).
#statistics #mathmodel
[Русская версия]
Our estimations on 22.12.2022 based on the model we published:
- killed in action: 127,823 combatants (+5,869 since the past week);
- captured: 27,603 combatants (-201);
- active AFU personnel: 325,241 combatants (-12,408).
Project statistics:
- the number of confirmed records KIA: 5,650 (+209); captured: 5,988 (+94);
- search requests received: 4,856 (+208) (multiple requests on the same person are counted as a single request);
- for 1,316 (+62) requests we provided information,
- probability to find something upon request: 27.1% (+0.1%, roughly 1 in 4).
#statistics #mathmodel
Our estimations on 22.12.2022 based on the model we published:
- killed in action: 127,823 combatants (+5,869 since the past week);
- captured: 27,603 combatants (-201);
- active AFU personnel: 325,241 combatants (-12,408).
Project statistics:
- the number of confirmed records KIA: 5,650 (+209); captured: 5,988 (+94);
- search requests received: 4,856 (+208) (multiple requests on the same person are counted as a single request);
- for 1,316 (+62) requests we provided information,
- probability to find something upon request: 27.1% (+0.1%, roughly 1 in 4).
#statistics #mathmodel
[Русская версия]
Our estimations on 29.12.2022 based on the model we published:
- killed in action: 134,444 combatants (+6,621 since the past week);
- captured: 28,027 combatants (-424);
- active AFU personnel: 322,533 combatants (-2,708).
Project statistics:
- the number of confirmed records KIA: 5,823 (+173); captured: 6,051 (+63);
- search requests received: 5,014 (+158) (multiple requests on the same person are counted as a single request);
- for 1,352 (+36) requests we provided information,
- probability to find something upon request: 26.9% (-0.2%, roughly 1 in 4).
#statistics #mathmodel
Our estimations on 29.12.2022 based on the model we published:
- killed in action: 134,444 combatants (+6,621 since the past week);
- captured: 28,027 combatants (-424);
- active AFU personnel: 322,533 combatants (-2,708).
Project statistics:
- the number of confirmed records KIA: 5,823 (+173); captured: 6,051 (+63);
- search requests received: 5,014 (+158) (multiple requests on the same person are counted as a single request);
- for 1,352 (+36) requests we provided information,
- probability to find something upon request: 26.9% (-0.2%, roughly 1 in 4).
#statistics #mathmodel