XPost: alt.paranet.ufo   
      
   Executive Summary:   
   - We split animal mass deaths up between the countries where they have   
     been reported.   
   - We determine which "weather" data and which keywords in UFO   
     sightings best predict mass animal deaths in each country.   
   - We order countries by the proportion of predicting factors that are   
     UFO-related (versus weather related).  If animal mass deaths are a   
     proxy for how much UFO activity is happening in each country then we   
     should be interested in which countries are seeing more and which   
     less activity.   
   - We find the list of national demographics -- drawn from the complete   
     list of from a couple different encyclopedias -- that predict the   
     ordering above suggests "UFOs" are "interested in" countries that   
     resemble Scandinavian countries. Perhaps this reflects something   
     about UFO's themselves.   
   - Has anyone checked with the Swedish Air Force to find out if they   
     have a solid alibi for 1930-2020?   
      
      
   We've seen that UFO activity seems to be linked with certain animal   
   mass deaths, in particular sea and lake animals.   
      
   We established the link to a statistical certainty (99%) by building   
   validated predictive models for different groups of animal deaths and   
   showing that adding in UFO sighting information improved the model and   
   a statistical test verified the information from the sighting data   
   "could not be ignored". The models were constructed from a large   
   corpus of mostly physical data from satellites aka "weather data" in   
   such a way the model was "optimal". The boost from adding sighting   
   data showed some information inside that data was needed to further   
   explain (in a statistical sense) why mass animal deaths were happening.   
      
   But the s/w that did all that work also proceeded to find "patterns in   
   the patterns". It asked itself how many weather variables were   
   relevant to various mass deaths, versus how many "features" relevant   
   to UFO activity were relevant.   
      
   In a series of experiments it tried to build models predicting mass   
   deaths in various countries from weather data and keywords found in   
   the description of UFO sightings.   
      
   E.g. for India it found the mass death database had 93 entries from   
   2011 to 2019. It found the best 20 correlates with the India mass   
   animal death data were:   
      
   variable/keyword	R2 against India's mass-death data   
   gavdiffgoing	 	0.209667	*   
   world-30	 	0.168622   
   gavdifffast	 	0.150387	*   
   world-20	 	0.150253   
   gavuah_globe6NoPol	0.146567   
   gavcntyellow	 	0.13853		*   
   arc-30	 		0.134785   
   gavlat-10	 	0.133826   
   arc-40	 		0.132713   
   gavworld-40		0.132666   
   cntwhite		0.132229	*   
   gavstorms-50		0.132039   
   world0	 		0.1296   
   arc0	 		0.128071   
   cntirregular		0.12483		*   
   world-50		0.122995   
   russia	 		0.122629   
   gavstormseg-150		0.122367   
   canada	 		0.121937   
   gavworld-120	 	0.120564   
      
   The naming of variables/keywords is somewhat inscrutable given it's   
   selected by the s/w using its down criteria. But the first line is   
   related to UFO sightings that contain the word "going".  It seems of   
   single variables whether weather-related or UFO sighting-related it is   
   the best, predicting about 21% of India's animal mass deaths.   
      
   The 2nd line is the AI's name for "avg monthly sea surface   
   temperatures down longitude 25W+-5deg". It finds this "weather" data   
   predicts about 17% of India's animal mass deaths.   
      
   Likewise the other variables are either weather variables or   
   "features" derived from keywords in UFO sighting data for each month.   
   I've marked lines corresponding with UFO sighting features with a (*).   
   So it seems out of the top20 simple models for India's animal deaths   
   5/20 are related to keywords in UFO sightings for the relevant months.   
      
   We can proceed for each country where animal deaths have been noted   
   and compile the top20 variables (as above) and determine how many   
   relate to UFO's.   
      
   We will then have a new dataset relating countries to a number   
   representing "how many ways" UFO sightings "interact" with country   
   mass deaths.   
      
   If we do that we get this data table:   
      
   Country		#UFO keywords in top 20 simple models for mass deaths   
   		in that country   
   argentina	7   
   australia	7   
   brazil		18   
   bulgaria	15   
   canada		16   
   chile		8   
   china		13   
   colombia	16   
   france		16   
   germany		17   
   greece		13   
   india		5   
   indonesia	16   
   ireland		17   
   italy		14   
   japan		9   
   malaysia	10   
   mexico		7   
   nepal		19   
   netherlands	14   
   norway		9   
   peru		15   
   philippines	5   
   russia		12   
   s_korea		18   
   sweden		14   
   vietnam		9   
      
   This looks like a total irrelevant jumble, but it actually contains   
   very interesting information, no matter how noisy.   
      
   We can ask -- "what national characteristics are most similar to this   
   table of ``UFO keyword counts''?".   
      
   We might posit that the more UFO features displace weather variables   
   the more UFO's "interact" with animals in each country.  Perhaps it is   
   a proxy for the number of sightings that SHOULD be reported from each   
   country, if only those countries were seriously collecting and   
   publishing that data.   
      
   It also might give us an idea what the UFO's are doing.  Why do they   
   fly more often over some countries than others?  It's been speculated,   
   e.g., UFO's are "interested in nuclear power", or one thing or another.   
      
   Luckily we have a program that can search ~9000 data sets and figure   
   out which national demographics "looks like" the above table.   
      
   Here's that list:   
      
   		Correlation with above "UFO links" table.   
   Demographic	R2		Beta   
      
   ren		0.683946	0.162265   
   femwork		0.593184	0.364934   
   urban		0.532009	-0.236026   
   manuf		0.45975		0.681975   
   unemp		0.447461	0.883491   
   hh		0.436024	-2.07093   
   gdpcap		0.402594	0.00023103   
   colcap		0.37143		-0.0335775   
   tothel		0.35881		1.39157   
   coalcp		0.351187	-0.0344364   
   prihel		0.349597	3.69688   
   pricon		0.34589		-0.261624   
   saltcp		0.330953	-0.0108056   
   murders		0.318431	-0.435983   
   prheex		0.303282	2.86779   
   ag		0.298916	-0.15751   
   infectdi	0.28345		-1.46826   
   wind		0.280375	-1.87441e-05   
   zfert		0.271938	-1.98775   
   PSUI		0.269113	-0.211441   
   zunder15	0.268728	-0.187557   
   wmuncp		0.261356	-0.0238411   
   comin5		0.2609		5.83367   
   fordebt		0.246415	-0.0858908   
   impcap		0.245439	0.000524649   
   trdcap		0.241422	0.000245041   
   incgdp		0.239886	0.233291   
   wmun		0.23891		-0.00011705   
   colpro		0.230377	-1.3297e-05   
   wagcp		0.22799		0.000642778   
      
   From this 2nd table we can see line 1 says variable "ren" seems to be   
   68% like the UFO links table. "Ren" is the% of renewable power   
   in a country around 2000. The beta column predicts for each 1 pct   
   point of renewables in a country there will be 16% more UFO keywords   
   in the top 20 that explain mass animal deaths than weather variables.   
      
   It seems the Number One criteria UFO's have for flying over a country   
   is whether or not it has a lot of renewable power!   
      
   In line 2 we find "femwork". This is the% of females in the   
   labor force. The variable explains 60% of the UFO links table.  For   
   each pct point of women in the workforce 36% more UFO keywords are to   
   be found in the list of top 20 explanatory variables for mass animal   
   deaths in that country.   
      
   It seems the Number Two reason a UFO has with flying over a country   
   and bumping into its animals is whether or not there is equal   
   opportunity for female employment.   
      
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)