XPost: alt.ufo.reports   
      
     If you give me someone's credit card records I can tell you where   
     and when the holder will buy something, what it will be and how much   
     they'll spend on it.   
      
   EXECUTIVE SUMMARY:   
   - Comparing UFO sighting data and earthquake data suggests where the   
     phenomenon might be based and the geographic regions it tends to   
     travel to and from.   
   - Certain regions seem to be the "source" of some significant part of   
     the traffic whereas others suggest "destinations".  Pinning down   
     these regions might be key to determining "what they are   
     doing". Although preliminary data suggests "farming" with maybe a   
     spot of abduction.   
   - Some of the data suggests a degree of "predictive ability" for   
     UFO's. They seem able to judge some characteristics of further   
     quakes and less able to judge others.  "Predictive ability" that   
     appears too might might correspond with some degree of causing the   
     related phenomena.   
      
      
   It seems possible to predict from where and to where UFO's are flying   
   based on a "shake the tree and see what falls out" principle.   
      
   In this report we'll use earthquakes of mag >= 5 to supply a realistic   
   shaking. I'll use the data 2006-2020 from from GEOFON GFZ Potsdam for   
   timing and location of the shaking and NUFORC's public sighting data   
   for the falling out of trees part of the relationship.   
      
   Here we'll only look at the general form of the results.  Later we'll   
   break it down by UFO type (usu shape and/or color) and even later try   
   to pin down specific geographic locations (within 1 mn km2, anyway :)   
   that seem to be the focus of "source" and "destination" traffic.   
      
   I'll again rely on some s/w to perform robust analysis, handle some of   
   the reasoning about what could possibly be the cause of what, and dig   
   out the most interesting results for a large number of calculations.   
      
   The procedure will involve comparing the "shape" of the data against   
   the UFO sighting record and finding where they are too similar to be a   
   fluke. We will look at the location of groups of earthquakes at the   
   resolution of 1 month and 10 deg bands of latitude and longitude, as   
   well as the world-wide averages for quake latitude, longitude and depth.   
      
   For each data-set we'll try various strategies to bring out higher   
   degrees of similarly between the "stimulus" (quake) and "response"   
   (sighting) data. Some of these transformations are seemingly bizarre   
   but that's what the AI s/w is for -- to find bizarre things that it   
   knows from wide experience will "work" even if they would not be   
   thought of by a human being.   
      
   We'll also use a limited amount of time shifting to look for   
   matches. We can move the "stimulus" back and forward by up to +-12   
   months to look for a match. -ve shift are matching later stimulus   
   against earlier response and can illustrate -- in the case of a   
   rational phenomenon -- a degree of anticipation. In this case, perhaps   
   UFO's can determine up to a year in advance whether a moderately large   
   cluster of earthquakes will occur and take appropriate steps.   
      
   In the case of +ve time shifts we are comparing earlier stimulus   
   against later response and underlining whether there is some delay in   
   the test animal's response in interpreting the stimulus and acting on it.   
      
   It turns out there are very few stimulus/response patterns that are   
   statistically significant.   
      
   The summary results are:   
      
   Quake Data	Lag		R2		beta		90% CI   
   		(m)   
   sqrtqdepth	5		 0.818724	-75.7413 	14.1379   
   sqrtqcounts	-7		 0.437159	-22.8378 	8.17623   
   sqrtqlat	12		 0.302935	19.2428 	8.7406   
   sqrtqlng	-10		 0.20577	-11.0352 	5.03491   
      
   All of these results are extremely robust -- essentially tested at 1%   
   significance. IOW there is only about 1 chance in 100 any one of them   
   might be due to just some lucky coincidence in the data.  There is   
   therefore about 95% chance all 4 of them together are non-chance.   
      
   If you plot any of them they look like points scattered very close to a   
   straight line. Note all the data are subject to a SQRT transform.  (In   
   the case of latitude and longitude is a "signed sqrt" that transfers   
   any -ve sign of the argument to the final value of the sqrt).   
      
   The top result -- where the R2 says the data is least noisy and the   
   relative error in the \beta is least -- says deeper quakes seem to   
   result 5m later in fewer UFO sightings.  This suggests UFO's could be   
   based quite deep under the earth.  A big disturbance down there might   
   cause them to "lock down".  Remember, this is the overall pattern   
   world-wide and may be (later) contradicted by the situation in specific   
   latitude or longitude bands.   
      
   The 2nd line says more earthquakes are related to lower UFO   
   sightings. But -- unusually -- the Lag indicates 7m of anticipation   
   is involved. Apparently the reduction in UFO sightings happens 7m   
   before the earthquakes actually happen. This suggests 2 possibilities --   
   (a) UFO's can predict earthquake clusters quite far out,   
   (b) all that underground base building make be *causing* at least   
   some earthquake activity.   
      
   #3 relates the "average latitude" of earthquakes in a given month   
   with UFO activity/sightings. The centroid of monthly quake activity   
   moves a little between the N and S hem -- generally only up to 10 deg   
   from the equator. But it seems the further N the center of quake   
   activity is, the more UFO sightings are to be expected about 12m   
   later. The further S quake activity moves the less UFO activity is   
   expected.   
      
   Remember "UFO activity" is a shorthand for "Sightings in the NUFORC   
   database" which we might take to be a proxy of global UFO activity.   
      
   My interpretation of "more activity" and "less activity" with   
   respect to earthquakes is as follows:   
   - More activity suggests an quake is striking close to home and the   
     UFO's have to temporarily vacate the region.  When a big shake hits   
     LA a lot of people run out into the streets or drive into the desert.   
   - Less activity suggests quakes are striking in some region UFO's   
     generally fly to for work or recreation. The more quakes that hit   
     Disney World the less patronage they might expect.   
      
   So together these considerations suggests UFO's live mostly   
   underground somewhere in the N Hem and seem to regularly travel to   
   other locations in the S Hem for reasons unknown unless you look at   
   the plankton data.   
      
   Finally, the last line suggests there is a similar relationship   
   between longitude and sightings as with latitude.  More quakes in the   
   W Hem are associated with more UFO activity; more quakes in the W Hem   
   are associated with less UFO activity.  Interestingly, this   
   relationship has a -ve lag -- meaning the UFO's may be planning their   
   E/W distribution 10m in advance because they have some idea when and   
   where quakes will be happening.   
      
   While the R2 is quite high for the "depth" relationship, they are moderate   
   or low for the others. This suggests for the -ve lags the UFO model   
   for quakes is more certain of the number than longitude of quake clusters.   
      
      
      
   In subsequent posts we'll narrow down some of the regions involved and   
   divide things up by UFO type to look at the different patterns that emerge.   
      
      
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)