Neural Networks and Correlation versus Causation

I need to prove that certain patterns were engineered and not random events and have been told that Neural Networks can be used to help make this determination.

What kinds of tools and algorithms should I be looking at?

Any sample projects that have done something similar?
newbiewebSr. Software EngineerAsked:
Who is Participating?
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

What sort of patterns?  Who would have engineered them?  Where do they come from?  What is the purpose of this proof?  

And who suggested I it was possible or necessary to prove anything about a pattern?

Neural networks can be trained to recognize patterns, but I don't see how they can be used to prove anything.
newbiewebSr. Software EngineerAuthor Commented:
I have created the largest database of 9/11 events in the world and included it in my book:

I have also found patterns built into the roads, buildings and parks in Boston, Harvard and Cambridge which were heralding the 9/11 flights out of Logan Airport, well before 2001. I correlate these patterns with the history of 9/11, which dates back to the ancient world.

So, I hope to have a mathematical basis to complement my book, which would show Causation and not Correlation.

Engineered to match occult patterns which date back to Babylonian times. For example, the earliest 9/11 attack I found was in the year 586 B.C. when Babylonian King Nebuchadnezzar destroyed the First Temple of Jerusalem (King Solomon's Temple) on the 9th day of the 11th month.

I have been told Neural Networks can be used when showing Correlation versus Causation by the same guy who told me to look into the NECSI, the New England Complex System's Institute, at MIT.
From Nebuchadnezzar to today is 2500 years, enough time for all kinds of events.

It would be good to have an a priori definition of the events of interest. Then you can find them all, perhaps assign some sort of weighting, and bin them by date. A statistical analysis could tell you which day of the year is most interesting.

If you only look at September 11th's, then what you find is coincidence by definition.  There is no mathematical tool or technique that could find or prove connection.

Wikipedia lists a different date for Solomon's Temple: August 28th.
Determine the Perfect Price for Your IT Services

Do you wonder if your IT business is truly profitable or if you should raise your prices? Learn how to calculate your overhead burden with our free interactive tool and use it to determine the right price for your IT services. Download your free eBook now!

Neural networks don't determine correlation vs correlation.
Causal networks can be used to eliminate causal models.
One caveat is that the larger the database you examine, the greater the chance of finding accidental correlations.

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
Maybe you should watch this Myles Power video:

I've shared your book link with Myles.  Perhaps he will make a video about your book.
newbiewebSr. Software EngineerAuthor Commented:
> Wikipedia lists a different date for Solomon's Temple: August 28th.

On the Hebrew Calendar in 586 B.C., the 9th Day of the 11th Month, the temple was destroyed. It is perhaps the most famous Jewish holiday and a day of deep mourning. Then, on the 9th Day of the 11th Month on the Hebrew Calendar in 70 A.D. Roman Emperor Titus destroyed the Second Temple of Jerusalem.

The month of Av was the 11th month.

Fast forward to the Gregorian Calendar.

Hitler's Coup was on the 9th Day of the 11th Month, in European date format: 9/11/1923

Hitler's attack on 30,000 Jews in Germany (Night of the Broken Glass) was on the 9th Day of the 11th Month, in European date format: 9/11/1938.

I KNOW these patterns prove Causation, because I also have researched the underlying occult connections which worship these dates. But I want to find a mathematical number that proves this.

> It would be good to have an a priori definition of the events of interest.

I do have that, but that is the longer version of the story.

For now, is there any way to get some level of confidence of Causation, short of me building some massive database of other dates?
newbiewebSr. Software EngineerAuthor Commented:
Ozo, nice video: Correlation CAN Imply Causation! | Statistics Misconceptions

I like the idea of eliminating the impossible. My approach has followed that of Sherlock Holmes, actually:

"Once you eliminate the impossible, whatever remains, no matter how improbable, must be the truth."

This approach combines the use of inductive and deductive logic, as the Yin and Yan, and has proven quite powerful for me.
What is the falsifiable statement of your thesis?  Does it have one?

How can you KNOW something BEFORE calculating a confidence level?

What are your numeralogical inputs, constraints, and error bars?
With Hebrew, Georgian, Julian, Mayan, Babylonian, ... calendars,
PlUS month/day,  day/month, min/hour, and hour/min, ...
PLUS longitude, latitude, street address, and ZIP code, ...
PLUS day of, day before, day after, day intended, ...

With enough slop on the inputs you can convince yourself of almost anything.

The first plane hit the WTC at almost exactly 9:11 on 9/11.  What are the odds?
newbiewebSr. Software EngineerAuthor Commented:
I need to have a falsifiable statement in my thesis, I agree. Only then can I get a numerical estimate to show it was not coincidental.

What might that look like?
This is probably a good place to start.
A pertinent snippet from the answer:

Martin Gardner uses the example of finding "fiveness" in the Washington Monument to show you can find significance in any numbers (Numerology) if you look hard enough. Here's a list of what's special about the first 10,000 numbers (mostly mathematical). Similarly, you can find something interesting about every day of the calendar year. Throw in Bible and Qur'an references for good measure, feel free to pick from any book/chapter. Use 9:11 or 11:9 or 911 or 119 or 9.11 or 11.9.
newbiewebSr. Software EngineerAuthor Commented:
>>   What might that  [a falsifiable statement in my thesis]  look like?

This is the art in science.  You have noticed some events.  Presumably you have noticed some relationship between events.  The next step is to formulate a testable hypothesis to confirm, refine, or repudiate your assumption.

So what is your assumption?  And it must remain an assumption until you test it.
  1. Things happen.
  2. Bad things happen.
  3. Bad things happen on 9/11 or 11/9.
  4. Bad things only happen on 9/11 or 11/9.
  5. More bad things happen on 9/11 or 11/9 than on any other pair of dates.
  6. Bad things will continue to happen on 9/11 or 11/9.
  7. There is evidence that allows me to make specific predictions about future 9/11 or 11/9 events.

1, 2, and 3 are testable and true, but are so general as to be uninteresting.
4 is probably false unless you have a strange definition of bad things.

5 is testable and interesting.  But you would have to start with careful definition of bad things and do an enormous amount of research.  
If you go back to 587 BC for supporting evidence, you have to do the same for the other dates as well.
  • What sort of events are you considering?  You have mentioned acts of terrorism, war, and repression.
  • Do you also include plagues, famine, floods, fires, earthquakes, volcanic eruptions, bridge and dam failures, and plane crashes.
  • Is there a threshold for deaths or property damage?
  • Is the local date convention binding? 9/11 in the US, and 11/9 in Europe?  Or does it depend?
  • Is there a filter for location and affected population?  Do Hiroshima and Nagasaki make the cut?
  • Can events last more than one day?  If so, do they have to start on the target date or just include it?

6 is probably true.  Testing is just waiting.  But it is too general to be useful or interesting.
newbiewebSr. Software EngineerAuthor Commented:
Thank you kindly for that outstanding answer to:

>   What might that  [a falsifiable statement in my thesis]  look like?
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today

From novice to tech pro — start learning today.