Newsgroups: alt.paranet.ufo
From: kymhorsell@gmail.com
Subject: how to predict mars quakes #2

[uploaded 33 times; last 28/09/2024]

EXECUTIVE SUMMARY
- We re-visit the InSight data on Mars quakes -- now updated to the
  end of 2022 (v14).
- We build 2 modest NN-based models (4000 coeff ea) to predict Mars quakes.  
  One based on the position of the planets (notably Jupiter; a major
  influence on Mars quakes) and 100s of asteroids that orbit between
  Mars and Jupiter. The 2nd based on sightings of various types of UFO
  including different colors, shapes, time-of-day seen, direction of
  motion, etc.
- Both models perform much better than guessing (i.e. using the daily
  average number of quakes). And both models are amazingly close to
  each other. It's as if the UFO sightings data "encodes" the
  position of all the planets and asteroids of the other model.
- As found in a prev post we have evidence to suspect UFO's are
  somehow connected with the position of planets and asteroids and may
  be implicated in the creation of some of the grinding and bumping
  noises heard by the InSight probe's seismometer.


In a prev post I looked at the data on Mars quakes being gathered by
the InSight robot. The data has been updated recently to run more or
less continuously between the start of 2020 and the end of 2022. 
Almost 15,000 tremors, jolts and odd noises have been recorded in that time.

An XML file with the dataset is available here:
<http://ds.iris.edu/files/insight/v14/events_extended_multiorigin_v14_2023-01-01.xml>.

If you want to reproduce what I've done the day by day count of events
looks like:

2020.000 53 2020.003 20 2020.005 32 2020.008 25 2020.011 32
2020.014 42 2020.016 20 2020.019 52 2020.022 55 2020.025 22
2020.027 18 2020.030 15 2020.033 45 2020.036 30 2020.038 43
2020.041 45 2020.044 77 2020.046 45 2020.049 79 2020.052 37
2020.055 65 2020.057 15 2020.060 56 2020.063 20 2020.066 33
2020.068 39 2020.071 15 2020.071 5 2020.074 42 2020.077 32
2020.079 37 2020.082 48 2020.085 35 2020.087 64 2020.090 31
2020.093 53 2020.096 48 2020.098 10 2020.101 10 2020.104 35
2020.107 58 2020.109 44 2020.112 82 2020.115 42 2020.117 28
2020.120 55 2020.123 25 2020.126 33 2020.128 38 2020.131 44
2020.134 25 2020.137 32 2020.139 15 2020.142 28 2020.145 30
2020.148 38 2020.150 28 2020.153 20 2020.156 27 2020.158 15
2020.161 15 2020.164 27 2020.167 25 2020.169 28 2020.172 10
2020.175 10 2020.178 33 2020.180 20 2020.183 5 2020.186 15
2020.189 10 2020.191 52 2020.194 37 2020.197 20 2020.199 38
2020.202 31 2020.205 5 2020.208 5 2020.210 5 2020.213 21
2020.216 32 2020.219 20 2020.221 10 2020.224 20 2020.227 20
2020.230 22 2020.232 20 2020.235 5 2020.238 42 2020.240 27
2020.243 14 2020.246 28 2020.249 10 2020.251 20 2020.254 30
2020.257 10 2020.260 20 2020.262 10 2020.265 51 2020.268 5
2020.270 5 2020.273 10 2020.276 10 2020.279 5 2020.281 22
2020.284 10 2020.287 18 2020.290 23 2020.292 22 2020.295 15
2020.298 5 2020.301 15 2020.303 10 2020.306 5 2020.309 37
2020.311 5 2020.320 10 2020.322 5 2020.325 5 2020.328 15
2020.331 26 2020.336 5 2020.339 17 2020.342 5 2020.347 10
2020.358 12 2020.361 5 2020.363 5 2020.369 5 2020.377 5
2020.383 5 2020.402 14 2020.404 12 2020.426 12 2020.484 8
2020.489 5 2020.500 8 2020.503 5 2020.738 12 2020.760 12
2020.781 5 2020.792 12 2020.902 13 2020.904 12 2020.907 12
2020.951 14 2020.964 12 2020.984 5 2021.000 5 2021.005 13
2021.008 5 2021.011 5 2021.014 5 2021.027 12 2021.027 27
2021.036 17 2021.036 41 2021.046 43 2021.046 5 2021.049 1
2021.049 33 2021.052 11 2021.057 20 2021.057 5 2021.060 5
2021.060 74 2021.079 12 2021.098 10 2021.107 25 2021.112 12
2021.128 12 2021.131 14 2021.134 12 2021.137 5 2021.139 5
2021.150 17 2021.156 5 2021.158 24 2021.172 5 2021.178 22
2021.183 5 2021.186 5 2021.194 13 2021.197 15 2021.199 8
2021.202 10 2021.208 32 2021.216 10 2021.219 10 2021.221 10
2021.224 13 2021.227 17 2021.230 10 2021.232 12 2021.235 17
2021.238 12 2021.240 29 2021.243 21 2021.246 10 2021.249 20
2021.251 10 2021.254 24 2021.257 5 2021.260 5 2021.262 13
2021.265 13 2021.268 17 2021.273 8 2021.276 17 2021.290 12
2021.292 28 2021.298 5 2021.301 18 2021.303 10 2021.306 5
2021.309 5 2021.311 23 2021.314 8 2021.317 5 2021.320 17
2021.322 18 2021.325 10 2021.331 53 2021.333 26 2021.336 25
2021.339 21 2021.342 29 2021.344 44 2021.347 29 2021.350 20
2021.352 29 2021.355 13 2021.358 12 2021.361 13 2021.363 25
2021.366 25 2021.369 18 2021.372 22 2021.374 24 2021.377 8
2021.380 34 2021.383 30 2021.385 41 2021.391 13 2021.393 37
2021.396 30 2021.399 30 2021.402 38 2021.404 30 2021.407 29
2021.410 13 2021.413 38 2021.415 18 2021.421 30 2021.423 10
2021.426 18 2021.429 62 2021.432 26 2021.434 26 2021.437 29
2021.440 25 2021.443 21 2021.445 53 2021.448 41 2021.451 38
2021.454 23 2021.456 55 2021.459 38 2021.462 40 2021.464 35
2021.467 30 2021.470 17 2021.473 70 2021.475 29 2021.478 77
2021.481 26 2021.484 49 2021.486 40 2021.489 45 2021.492 78
2021.495 41 2021.497 80 2021.500 42 2021.503 45 2021.505 42
2021.508 41 2021.511 33 2021.514 46 2021.516 50 2021.519 54
2021.522 25 2021.525 49 2021.527 30 2021.530 58 2021.533 29
2021.536 41 2021.538 25 2021.541 24 2021.544 25 2021.546 29
2021.549 33 2021.552 58 2021.555 42 2021.557 18 2021.560 28
2021.563 33 2021.566 9 2021.568 46 2021.571 30 2021.574 38
2021.577 37 2021.579 41 2021.582 24 2021.585 37 2021.587 42
2021.590 58 2021.593 50 2021.596 29 2021.598 33 2021.601 34
2021.604 25 2021.607 25 2021.609 30 2021.612 20 2021.615 24
2021.617 21 2021.620 29 2021.623 37 2021.626 28 2021.628 13
2021.631 29 2021.634 29 2021.637 36 2021.639 48 2021.642 45
2021.645 55 2021.648 17 2021.650 32 2021.653 39 2021.656 17
2021.658 25 2021.661 27 2021.664 45 2021.667 20 2021.669 33
2021.672 22 2021.675 56 2021.678 57 2021.680 42 2021.683 33
2021.686 42 2021.689 60 2021.691 37 2021.694 33 2021.697 85
2021.699 10 2021.702 10 2021.705 33 2021.708 26 2021.710 82
2021.713 40 2021.716 17 2021.719 10 2021.721 40 2021.724 25
2021.727 46 2021.730 17 2021.732 57 2021.735 29 2021.738 50
2021.740 25 2021.743 42 2021.746 41 2021.749 62 2021.751 18
2021.754 62 2021.757 45 2021.760 33 2021.762 37 2021.765 33
2021.768 33 2021.770 44 2021.773 40 2021.776 54 2021.779 37
2021.781 49 2021.784 53 2021.787 57 2021.790 67 2021.792 50
2021.795 17 2021.798 24 2021.801 43 2021.803 47 2021.806 37
2021.809 70 2021.811 62 2021.814 34 2021.817 44 2021.820 38
2021.822 29 2021.825 49 2021.828 18 2021.833 35 2021.836 64
2021.842 41 2021.844 74 2021.847 83 2021.855 47 2021.858 18
2021.861 26 2021.863 22 2021.866 29 2021.869 45 2021.872 53
2021.874 36 2021.877 37 2021.880 42 2021.883 55 2021.885 50
2021.888 51 2021.891 48 2021.893 15 2021.896 47 2021.899 5
2021.902 69 2021.904 68 2021.907 34 2021.910 18 2021.913 60
2021.921 51 2021.923 61 2021.926 95 2021.929 42 2021.932 42
2021.934 50 2021.937 91 2021.940 45 2021.943 47 2021.945 30
2021.948 49 2021.951 36 2021.954 69 2021.956 42 2021.959 46
2021.962 48 2021.964 54 2021.967 52 2021.970 45 2021.973 23
2021.975 100 2021.978 26 2021.981 50 2021.984 52 2021.986 53
2021.989 47 2021.992 15 2021.995 35 2022.000 27 2022.003 54
2022.005 36 2022.008 48 2022.011 53 2022.014 22 2022.079 32
2022.082 13 2022.085 46 2022.087 10 2022.090 71 2022.093 44
2022.096 61 2022.098 17 2022.101 37 2022.104 21 2022.107 10
2022.109 10 2022.115 10 2022.117 22 2022.120 18 2022.123 5
2022.126 13 2022.128 15 2022.131 25 2022.134 22 2022.137 5
2022.139 25 2022.142 5 2022.145 34 2022.148 10 2022.150 15
2022.153 21 2022.156 64 2022.161 5 2022.164 19 2022.167 5
2022.169 5 2022.172 13 2022.175 12 2022.180 12 2022.183 17
2022.189 12 2022.194 10 2022.197 37 2022.199 5 2022.202 5
2022.205 10 2022.208 13 2022.210 5 2022.213 8 2022.216 5
2022.221 5 2022.224 5 2022.232 8 2022.235 5 2022.238 5
2022.240 5 2022.254 17 2022.257 8 2022.268 20 2022.336 52
2022.339 8 2022.342 12 2022.374 12 2022.380 20 2022.661 12
2022.880 19

My s/w has been updated quite a bit since we last tried to predict
these events using various kind of other time-series.  This time I'll
present 2 new models -- one based on the position of the planets and
~500 asteroids over the period 2020-22 and the other model will use
the daily UFO sightings of ~500 different types of object (e.g. the
NUFORC-assigned "shape", the "color" as described in the comment field
of the short report, plus several other keywords describing direction
of motion and select details related to the type of light the object
was supposedly emitting).

As with other methods I've talked about the new s/w uses a "hold out"
method to validate the model created. 1/2 the data is held aside and
the model trained up on just the other 1/2. At the end the hidden data
is dragged out and used to measure how well the model works on data it
never saw before. We have evidence to believe the model captures
something about the real world if it can generalise well enough from
the training data how the target variable behaves even in new
conditions.

The underlying modelling method used here is a neural net (NN).  The
methodology was developed over the past 50y from modest beginnings that
included several spectacular but well-documented failings
(e.g. spotting tanks concealed in vegetation). But the kinks have been
worked out and NN of one kind or another (e.g. one example is the
"deep learning" methods you may have seen mentioned in newspapers) are
a staple for predictive modelling nowadays. Basically a NN takes all
the possible data that MIGHT help predict some target variable and
tries to manipulate it in "all possible ways" (or just a very large
number of them) to find the way that predicts the target best.  As I
mentioned, it generally uses a validation technique like withholding
some part of the dataset to use in the final stage to see how well the
model behaves with unseen data.

In the Mars quake data, above, the yardstick for how well the model
fits the data is given by the standard deviation of the day to day
events.  In this case that's 16.6. It we were to "predict" tomorrow's
number of Mars quakes we could at the very least use the daily average
of ~27.  If we did then the "average error" would be 16.6 -- the std
dev.  We hope the models the NN builder comes up with have a much
better error rate than that -- otherwise there is no point in using
them to predict Mars quakes because we could just use the daily
average, instead.

Amazingly (as far as I'm concerned :) the prediction error using the
planetary and asteroid positions is 12.2. This may not sound like a
big improvement over "just guessing" and using the average.  But every
point that's shaved off the error is generally regarded as a big
deal. In industry even a 0.1 improvement over guessing means money in
the bank and that's why data scientists command million dollar salaries.

The plot of the dataset and this first model is here:
<kym.massbus.org/MARS-QUAKES/SWITCH-NN/FINDBESTVALREGR/oe.gif>.  
You can see it doesn't simply predict a smooth curve that runs between
the points -- it tries to follow the data up and down and manages to
come fairly close to the actual values most of the time.  The blue
line at the bottom on the lhs shows which part of the data was used to
train the NN. The rhs of the plot is the test part.  It's hard to
imagine how anything could predict the rhs after only seeing the
lhs. But this is the "predicting the elephant's trunk" scenario that
AI regularly can do these days. Given just a picture of the animals
bottom AI algorithms can predict very closely what the other end
looks like even when it has totally unexpected (to humans) features.

So it seems we can predict about 2/3 of Mars quakes knowing the
positions of the planets and selected asteroids.  This is pretty
amazing because the positions of planets can be churned out on a
calculator many years into the future.  It's as if that 67% of Mars
quakes is "inevitable" and just part of the furniture in our solar
system. It doesn't matter how chaotic quakes may appear, they can be
predicted in many cases.  An interesting feature of chaos is that
prediction is possible.  A chaotic pattern -- different from a random
pattern -- has the quality of self-similarity. One part of the
pattern is "similar" (but not identical to) other parts of the
pattern. If you know one part of the pattern then you know within
limits what another part of the pattern looks like. This, in fact, is
why science works -- because the universe is in chaos.

Now what about the 2nd model, where we use data on the different kinds
of UFO sightings seen day by day over the same period?

It also is much better than just guessing. SOMEHOW the information of
Mars quakes is "encoded" in the UFO data. Knowing the numbers for all
the different UFO types seen on a particular day enables us to
calculate the number of expected Mars quakes to an avg error of
+-12.3.  Just a hair outside the position of all the planets and
asteroids from the first model.  It's hard to escape the implication
that UFO sightings "encode" the position of a lot of asteroids and
planets. It's as if they were connected. :)

The plot of the model fitting UFO sightings to daily Mars quakes is here:
<kym.massbus.org/MARS-QUAKES/SWITCH-NN2/FINDBESTVALREGR/oe.gif>.
Again, the lhs of the plot shows the part of the data used to train
the NN. The rhs shows the "test area". And, again, it's hard to see
how a NN could guess the shape of the rhs from the lhs.

--
"Nothing in life is to be feared, it is only to be understood.
Now is the time to understand more, so that we may fear less."
- Marie Curie

[Secret UFO recovery program blown open:]
I hope this revelation serves as an ontological shock sociologically
and provides a generally uniting issue for nations of the world to
re-assess their priorities.
-- David Grusch, 05 Jun 2023
[Talking to Les Kean et al for The Debrief, Grusch called for an end to
nearly a century of global UFO secrecy and warned that humanity needed to
prepare itself for "an unexpected, non-human intelligence contact scenario"].

[David Grusch's] assertion concerning the existence of a terrestrial arms
race occurring sub-rosa over the past eighty years focused on reverse
engineering technologies of unknown origin is fundamentally correct, as
is the indisputable realization that at least some of these technologies
of unknown origin derive from non-human intelligence.
-- Col Karl Nell (ret), 06 Jun 2023

Senator Admits UFO Whistleblower Report Tracks With Official Briefing
Hosted on MSN, 21 Jun 2023 15:44Z
Senator Josh Hawley said a recent UFO whistleblower's report is
"pretty close" to information he received in a briefing after the ...
[Pretty embarrassing. The Pentagon had denied most of Grusch's story
about captured tech and dead bodies; now it turns out they had told
Senators much the same stuff in secret briefings].

But what is true and I'm actually being serious here, is there are, there's
footage and records of objects in the skies that we don't know exactly what
they are, We can't explain how they moved, their trajectory. They did not
have an easily explainable pattern.
-- Pres Barack Obama, "The Late Show", 2021

"I think some of the phenomena we're going to be seeing continues to be
unexplained and might, in fact, be some type of phenomenon that is the
result of something that we don't yet understand."
--Ex-CIA Director John Brennan

"[F]or the few cases in all domains--space, air, and sea--that do
demonstrate potentially anomalous characteristics, AARO exists to help the
DOD, IC, and interagency resolve those anomalous cases. In doing so, AARO is
approaching these cases with the highest level of objectivity and analytic
rigor. This includes physically testing and employing modeling and
simulation to validate our analyses and underlying theories, and
peer-reviewing those results within the U.S. government, industry partners,
and appropriately cleared academic institutions before reaching any
conclusions."
-- Dr Sean Kirkpatrick, Senate Hearings on UFOs, 19 Apr 2023.