Merged Discussion of femr's video data analysis

Dr Crockett Grabbe - "Peer-reviewed article in Journal of Engineering Mechanics"

Meaningless technobabble.

The kind of thirty-second notes extracted from Charlie Parker solos are well Hertz in nature.

And they contain some noise. Beautiful noise.
 
Last edited:
Meaningless technobabble.

Incorrect.

As I said...

The kind of motion already extracted from videos is well sub-pixel in nature, and contains some noise.

The result of determining full three dimensional position (in other words correcting for perspective) for the trace data in question would be very minor, and an almost linear deskew. As the fidelity of the data is already sub-pixel, and the Dan Rather viewpoint is almost orthogonal, the resultant change to derived acceleration trend plots would also be minor. Perhaps a slight change in magnitude, and that's about it.

If you don't understand, take offers to explain to you in more detail.

Calling my statement "Meaningless technobabble" does you no favours.

Tagging onto tfk's nonsense is also a bad idea.
 
Can you suggest how many samples should I consider for taking as static reference? Maybe 400?
Given the significant negative effect the Audacity audio editing package noise removal algorithms have had upon the data during the actual descent, I would not recommend progressing.
 
Dr Crockett Grabbe - "Peer-reviewed article in Journal of Engineering Mechanics"

More technobabble. Good luck with this.
 
Savitzky-Golay smoothing, part 1

On several occasions, I have said I believe femr2's position/displacement data are better than NIST's. I have also said I am unimpressed by femr2's analyses of that data.

Within the past week or so, in other threads, femr2 has said several things that make it easier to explain my opinion of his analyses.

On 2 May 2011, femr2 posted an acceleration curve that he said had been obtained via Savitzky-Golay smoothing/differentiation of femr2's position/displacement data for the NW corner of WTC7.

Within hours, I pointed out that femr2 had once again failed to describe his methods and parameters with enough detail for others to duplicate his Savitzky-Golay acceleration curve. In particular, he had specified neither the degree of his best-fit polynomial nor the number of points in the window. femr2 responded by saying "So? 2, 50."

As an answer to my direct question, that 50 didn't make sense. With the Savitzky-Golay algorithm, as described in their paper, that window must contain an odd number of points. 50 is even, not odd. When I pointed out the problem, femr2 responded by saying "Possibly fair point" without resolving the problem.

That was more than a year and a half ago. femr2 still hasn't stated the parameters he used to generate that curve. To this day, we can only guess at femr2's parameters.

When I tried all plausible guesses as to what femr2 might have meant by "2, 50", I found that femr2's curve could not have been generated using the Savitzky-Golay algorithm with parameters that lie near the parameters stated by femr2.

The most plausible guesses as to what femr2 might have meant by saying "2, 50" were:
  • The best-fit polynomial approximation to the position/displacement data was of degree 2, and the width of the window was 50 (so there would have been 51 points in each window).
  • The best-fit polynomial approximation to the first derivative of the position/displacement data was of degree 2 (so the Savitzky-Golay polynomial would have been of degree 3), and there were 51 points in each window.
  • The best-fit polynomial approximation to the second derivative of the position/displacement data was of degree 2 (so the Savitzky-Golay polynomial would have been of degree 4), and there were 51 points in each window.
  • The half-width of the window was 50, so there were 101 points in each window, and the degree of the Savitzky-Golay best-fit polynomial was 2, 3, or 4 (as above).

The following graph compares the acceleration curves generated by the Savitzky-Golay smoothing/differentiation algorithm, with 51 points per window and polynomial degree n=2,3,4, against femr2's alleged Savitzky-Golay acceleration curve:

sg51.png



(Observant readers may notice that the green curve for degree n=2 doesn't show up in the graph. It's there, but it's covered up by the blue curve for degree n=3. Those two curves are so similar that their small differences aren't visible at the resolution of this graph. That's a meaningful observation, as explained below.)
​

From that graph, it's obvious that femr2 could not have generated his acceleration curve using the Savitzky-Golay algorithm with a window size close to 50.

Could 50 have been the half-width of his window? No:

sg101.png



(Once again, the green curve for degree n=2 is so close to the blue curve for degree n=3 that the green curve doesn't show up at this resolution.)
​

101 points per window gives us a rough approximation of femr2's acceleration curve, but it doesn't give us femr2's curve. The true Savitzky-Golay curve for n=3 and w=101 is considerably smoother than femr2's. Raising the degree to n=4 gives us a more nervous-looking curve, but not quite as nervous-looking as femr2's.

So those of us who tried to reproduce femr2's results have suspected for quite some time that he had either misapplied the Savitzky-Golay algorithm or had misreported the parameters he used.

Asking femr2 to describe his methods and parameters more clearly seemed pointless. After answering my request with the incorrect "2, 50", femr2 said he was "rather disinclined to acquiesce to your requests" for such details. Just a few days ago, femr2 clarified his meaning:

:) I said "If you asked nicely I may even repeat details about a specific graph, but with the ongoing derision and distortions that are going on I'm rather disinclined to acquiesce to your requests.".

"I'm rather disinclined to acquiesce to your requests." is a quote from Pirates of The Carribean ;)
"Means no."

By directing our attention to the piratical context of his phrase, femr2 has confirmed that the pirate's own two-word exegesis of his phrase speaks for femr2 as well.

Conclusion: femr2's failure to describe his methods and parameters in enough detail for others to reproduce his graphs has been deliberate.

None of this seemed terribly important, because low-degree Savitzky-Golay smoothing/differentiation with a window size on the order of one second or more does approximate femr2's acceleration curve:

sg61.png



(In this graph, I've stuck with best-fit polynomials of degree n=2. As can be seen in the other graphs, raising the degree to n=3 has very little effect on the graph. That tells us the additional features seen when the degree is raised still further, to n=4, are probably artifacts of the smoothing process interacting with noise in the signal. It might even tell us more, but I'll save that for later.)
​

The red curves in all three of the graphs above, labelled "femr2's iterated SG", were obtained directly from femr2's graph using software I wrote for that special purpose. That was the only way I could obtain a quantitative representation of femr2's acceleration curve, because neither I nor anyone else has been able to reproduce femr2's curve using the methods and parameters femr2 said he used.

Although we still don't know exactly how femr2 computed his acceleration curve, we now know he did not compute that second derivative using the algorithm described by Savitzky and Golay.

That too is a long story, so I'll save it for part 2.
 
Thanks for those graphs.


Is this the first set of graphs you have produced using SG or are there others from previous posts I have overlooked?




On the other thread I posted 3 examples of other researchers applying SG to the extraction of acceleration profiles from position data, just as is being done here. This is the better thread for these references and they are reproduced:



This is how the SG filter has been applied to the extraction of acceleration trend from position data in physics related studies by other people:



Application in physics:


http://www.google.com/url?sa=t&rct=...=dSW1Di0VUK-m2EumZg-S8Q&bvm=bv.42080656,d.aWc


http://www.google.com/url?sa=t&rct=...=G_bciIC80W5VIHXoiS1Q_w&bvm=bv.42080656,d.aWc


http://www.google.com/url?sa=t&rct=...=v6b8RgOp2wF-gZW692HI7Q&bvm=bv.42080656,d.aWc





>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>.

From the first link:

III.C. Computing Acceleration from Position via Numerical Differentiation
In order to determine acceleration from the videogrametrically-measured position data, we chose the approach of
differentiating the position data twice, using a simple numerical algorithm. While differentiation is an inherently
noisy operation, this approach has been used successfully in the past to compute velocities from experimentally-
determined position data.
For instance, Alexander and Colbourne [12] differentiated experimentally-determined
position data for several locations on a human limb segment, in order to obtain linear and angular velocities of the
limb. While no examples reporting the use of videogrammetric position data to obtain acceleration information
were found in the literature, there is no technical reason why this computation cannot be accomplished, as long
as appropriate numerical processing schemes are used to mitigate the noise amplified by the successive
differentiations.

V.A.ii. Determination of Acceleration via Numerical Differentiation of Smoothed Tracking Measurements
Prior work done by Kienle et al. [15] suggests that numerical filtering could be used to improve the results of
obtained when videogrammetry-based displacement data are twice differentiated to obtain accelerations.
Filtering prior (rather than subsequent) to differentiation was chosen because high-frequency noise components
caused by inherent measurement uncertainty tend to be significantly magnified by the differentiation process.




>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

From the second link:

The first step of the marker displacement data analysis consists in the time histories filtering using a
Savitzky-Golay polynomial filter (Sgfilter). Successively, after double derivative to calculate the acceleration, a
band Pass filter is applied to the acceleration time history. The choice of the most appropriate filtering parameters
has been made comparing the results from different combinations of the parameters. E.G. the figure 3 shows the
effect of the SGfilter with different polynomial degree and same windowing width.

>>>>>>>>>>>>>>>


W.D. Clinger, of those results in your last post, which one do you think most accurately depicts the acceleration trend of the NW corner of WTC7?
 
On 2 May 2011, femr2 posted an acceleration curve that he said had been obtained via Savitzky-Golay smoothing/differentiation of femr2's position/displacement data for the NW corner of WTC7.
Correct.

To this day, we can only guess at femr2's parameters.
Correct.

[qimg]http://www.cesura17.net/~will/Ephemera/Sept11/Femr2/sg51.png[/qimg]
My god, you've discovered colour !

From that graph, it's obvious that femr2 could not have generated his acceleration curve using the Savitzky-Golay algorithm with a window size close to 50.
Incorrect.

Conclusion: femr2's failure to describe his methods and parameters in enough detail for others to reproduce his graphs has been deliberate.
LMAO. I've certainly withheld detail from you Will, as firstly your attitude sucks, your constant tone of derision sucks, you firmly believe no-one can do things "right" except you, and it's funny as hell watching you speculate.

None of this seemed terribly important
Spot on son.

Although we still don't know exactly how femr2 computed his acceleration curve, we now know he did not compute that second derivative using the algorithm described by Savitzky and Golay.
Incorrect. You might want to consider whether SG was the only process applied, whether first and second passes used the same parameters, ... :D

That too is a long story, so I'll save it for part 2.
Yawn.
 
Soooo, what do we have now...

sg61.png


...and...

621711303.png


You know, I think it may have been said before, but...

I think I might see a trend emerging.

Would you Christmas Eve it eh.
 
More technobabble. Good luck with this.

femr likes technobabble, I don't know though whether he adopted this particular style of argumentation from the time when he supported the nanothermite camp or from the time when he supported the moon-landing deniers. (http://www.internationalskeptics.com/forums/showpost.php?p=8982600&postcount=48)

Having said that, there really is not much of a difference between femr's and W.D. Clinger's graphs.

But since some people have these weird Savitzky-Golay/acceleration profile fetishes, I guess this thread does have its right to exist.

:)
 
femr likes technobabble, I don't know though whether he adopted this particular style of argumentation from the time when he supported the nanothermite camp or from the time when he supported the moon-landing deniers. (http://www.internationalskeptics.com/forums/showpost.php?p=8982600&postcount=48)

Having said that, there really is not much of a difference between femr's and W.D. Clinger's graphs.

But since some people have these weird Savitzky-Golay/acceleration profile fetishes, I guess this thread does have its right to exist.

:)

Basically, toss measurement error in there and you get "you can't really be sure WHAT the acceleration is at any given point in time, however it looks like it went above g at least once and came back below it a few seconds later".

What does that tell you about the "MUST be because they blew up the COLUMNS 'cause it's FREEFALL[-ISH]" meme?
 
Basically, toss measurement error in there and you get "you can't really be sure WHAT the acceleration is at any given point in time, however it looks like it went above g at least once and came back below it a few seconds later"....
PLUS a second finding that there was evidence of movement for some time before the rapid descent stage occurred. Personally I find that second point more interesting than the "above G" finding - but I have the disadvantage of understanding the range of physics involved and I'm not tied down to the truther memes which link "G" to "must be CD".
...What does that tell you about the "MUST be because they blew up the COLUMNS 'cause it's FREEFALL[-ISH]" meme?
Yes - that is one aspect which also flows from the "second point" of "early movement".

Then another word for the GLOSSARY:
"technobabble" - a coded expression meaning either "I don't understand" OR "I don't want to understand" OR "I wish femr was not so explicitly clear as to his goal because I want him to have a different goal". :rolleyes:

Sort of "evaderbabble" OR "trollingbabble" but neither of those rolls of the tongue as good as techno..... :o
 
Last edited:
...
On the other thread I posted 3 examples of other researchers applying SG to the extraction of acceleration profiles from position data, just as is being done here. ...
No, it is not as done here. Did you read the papers you googled. They don't support using SG as being done here. You can't google up other studies and say they apply here. That is, just plain nonsense.

It is cute you think you can google your way to engineering excellence for a trivial study of wastestmaximustimeist.

...
This is how the SG filter has been applied to the extraction of acceleration trend from position data in physics related studies by other people:
You did not figure out why and what they used SG filter? Go ahead characterize the noise in each study and explain why and how they applied the SG filter. Good luck. ... you don't have clue what the filter is for.

Oh look, you found some physics stuff, so you think it applies.
You google up SG filter. Quote-mine what you think supports the SG used for this myopic study and fail to make a point. Just plain nonsense. Any progress on the search for the inside job? Did NIST aid the cover-up?
... some of the factors which, when studied in depth, show that the supposed "gravity-driven collapse" is a mere illusion to mask an intentional act so barbaric, so inhumane and morally impoverished that the fabled characteristics of Satan come to mind.

Please stay tuned as we discuss each of these factors in detail, while, of course, sticking to the subject originally posted in this thread.
Over 5 years and I have stay tuned and you have failed to discuss the factors or retract your failed inside job, mere illusion nonsense. How much longer? Or is the attack on others, the conclusions of your "book" the end all solution to 911?

Did you understand why they used the filter in the studies you googled? No - just plain nonsense.
 
In another thread there's been a discussion of SG with two passes (iterations), which is the method femr2 used to derive his acceleration plot with no theoretical backing, vs. using the second-order derivative of SG with one pass, which is the procedure suggested by Savitzky and Golay in their paper to derive acceleration data from signals with their smoothing method, as noted by W.D. Clinger.

At one point I asked:

Which one is better: 1 pass or 2 pass? [...]

ETA: And by the way, it would be interesting to see 1 pass SG overlaid with tfk's acceleration plot.
femr2's response:
Personally, I think my 2-pass curve is more useful than the 1-pass curve I posted, especially around the start of descent.
Well, so I went ahead and did the overlay. Here's the best graph I could come up with for 1-pass SG vs. 2-pass SG vs. tfk's version:

SG1vsSG2vsTFK.png


Now... what might be the basis for femr2 to say that the 2-pass curve is more useful? Based on what? It turns out that the 1-pass curve matches much better the trend that was arrived at independently by an experienced engineer by a different method. So I must logically ask, more useful for what purpose? Because the only purpose that comes to mind is self-deception. The match is a confirmation that femr2's two-pass method to derive the graph is objectively worse, in that it introduces deceiving artifacts.

I'd even go farther and say that this casts doubts on the technical merits of femr2's analyses.
 
......

LMAO. I've certainly withheld detail from you Will, as firstly your attitude sucks, your constant tone of derision sucks, you firmly believe no-one can do things "right" except you, and it's funny as hell watching you speculate.
.....

Yawn.


It doesn't seem femr2 is bored.
It sounds like he's sitting on a hot stove while typing.
 
PLUS a second finding that there was evidence of movement for some time before the rapid descent stage occurred. Personally I find that second point more interesting than the "above G" finding - but I have the disadvantage of understanding the range of physics involved and I'm not tied down to the truther memes which link "G" to "must be CD".
Yes - that is one aspect which also flows from the "second point" of "early movement".

Then another word for the GLOSSARY:
"technobabble" - a coded expression meaning either "I don't understand" OR "I don't want to understand" OR "I wish femr was not so explicitly clear as to his goal because I want him to have a different goal". :rolleyes:

Sort of "evaderbabble" OR "trollingbabble" but neither of those rolls of the tongue as good as techno..... :o



Authors and others who wish to convey a feeling of technical sophistication may write or talk in technobabble. They may use jargon without considering what it actually means to give an impression that they know things that their readers or listeners do not. However, if the jargon is decoded, it becomes apparent that the originator does not really understand what has been said or is deliberately being unclear. When used in this way, technobabble is considered pretentious and often unacceptable. If used inappropriately, even novice listeners can often detect that nonsense is being spouted forth.
 
Authors and others who wish to convey a feeling of technical sophistication may write or talk in technobabble. They may use jargon without considering what it actually means to give an impression that they know things that their readers or listeners do not. However, if the jargon is decoded, it becomes apparent that the originator does not really understand what has been said or is deliberately being unclear. When used in this way, technobabble is considered pretentious and often unacceptable. If used inappropriately, even novice listeners can often detect that nonsense is being spouted forth.
I'm aware of the accepted meanings. The point I made still stands - it is a pity those using "technobabble" here are not using the term legitimately. There is a serious topic for discussion - trivial false comments add nothing.

BTW what do you think of the "second point" I raised? It has far more significance than my throw away reference to the trivial game playing. :)
 
Savitzky-Golay smoothing, part 2

In part 1, I explained why femr2's alleged Savitzky-Golay acceleration curve fell under suspicion within hours of its introduction. Over the course of the next year and a half, femr2 was unable to describe his methods and parameters in enough detail for anyone to reproduce his graph. We now know femr2's failure was deliberate:

Conclusion: femr2's failure to describe his methods and parameters in enough detail for others to reproduce his graphs has been deliberate.
LMAO. I've certainly withheld detail from you Will, as firstly your attitude sucks, your constant tone of derision sucks, you firmly believe no-one can do things "right" except you, and it's funny as hell watching you speculate.


As consequence:

To this day, we can only guess at femr2's parameters.
Correct.


I also explained why I don't think this matters very much. femr2's acceleration graph does bear some resemblance to graphs produced by the Savitzky-Golay algorithm using somewhat wider windows than femr2 claims to have used. Since no one has ever been able to explain why femr2's oft-posted graphs matter, they probably don't.

Pointlessness is not a crime, and everyone should have a hobby.

Now that femr2 has confirmed the more speculative parts of part 1, we can defer further discussion of femr2's deliberately obstructive video analysis.

In this part 2, I will describe the Savitzky-Golay algorithm and how it fits into a much larger family of smoothing algorithms.

After all, I may not be the only person here whose pointless hobbies include learning more about smoothing and differentiation of time series.


[size=+1]The Savitzky-Golay Algorithm[/size]

In 1964, Abraham Savitzky and Marcel J E Golay published a paper that began by reviewing a variety of different methods for smoothing and differentiation of time series. They then confined their attention to one particular method (which is now known as the Savitzky-Golay algorithm, even though they didn't invent it), and described their clever technique for implementing that method (which is also known as the Savitzky-Golay algorithm).

To keep those two things straight, I'll refer to the method they implemented as Savitzky-Golay smoothing/differentiation, and I'll refer to their technique for implementing that method as the Savitzky-Golay implementation technique, or as Savitzky-Golay convolution.


[size=+1]Savitzky-Golay Smoothing/Differentiation[/size]

Savitzky-Golay smoothing is one member of the infinite family of smoothing methods that replaces each point of a time series by a smoothed point computed as follows:
  • Select a window (neighborhood) surrounding the point.
  • For some parameterized function (such as a polynomial of degree n), determine the parameters for which the function best fits the points within the window.
  • Using those parameters, calculate the value of the function at the point. That is the smoothed value.
Different methods use different windows, different kinds of parameterized functions, and different notions of best fit.

For each method of that family, the value of the kth derivative at each point can be estimated by computing the value of the kth derivative of the best-fit function at each point.

With Savitzky-Golay smoothing/differentiation:
  • The window is symmetrical about the point to be smoothed, with an odd number of points in the window.
  • The parameterized function is a polynomial of some degree n.
  • The best-fitting polynomial is defined as the one that minimizes the equally weighted sum of squared errors at each point within the window.
Savitzky-Golay smoothing/differentiation is popular because those are often good choices. Polynomials and their derivatives have obvious physical interpretations for position, velocity, and acceleration. Minimization of least squares is a familiar and convenient technique with pleasant mathematical and computational properties.

For many applications, the most questionable choice is the decision to give equal weight to the squared error at each point. Most methods give less weight to the error at points farther away from the center of the window.

Some of the engineers here would understand this better than I, but I am told that giving equal weight to the squared error at each point yields a fairly flat frequency response within a range determined by the width of the window.

On the other hand, this feature of Savitzky-Golay smoothing may give too much influence to outliers or noise at the extremes of a window. It also creates lag when wide windows are used to smooth a time series for position that may have resulted from sudden changes in acceleration.

That could be of concern when analyzing video of building collapses. To find out whether it's a problem, we'd have to understand the purposes of the analysis.


[size=+1]Savitzky-Golay Convolution[/size]

Although weighting the square error at each point equally may be questionable, it allowed Savitzky and Golay to implement the method more efficiently.

Savitzky and Golay noticed that finding the coefficients of the polynomial of degree n that minimizes the sum of squared errors over a window of width w is equivalent to computing n+1 convolutions, each of length w, using convolution coefficients that depend only on n and w.

These convolution coefficients can be computed ahead of time. For an example of how these convolution coefficients are derived, see Wikipedia's current article on numerical smoothing and differentiation.

All desired derivatives at the point can be computed directly from the smoothing polynomial's coefficients. That takes O(n) time for each derivative.

As an alternative, the desired derivatives can be computed via convolutions, with convolution coefficients calculated ahead of time as for the smoothing polynomial. That takes O(w) time for each derivative, so this alternative technique is competitive when windows are small.

The Savitzky-Golay paper contains tables showing the values of these convolution coefficients for windows with up to 25 points, smoothing polynomials of degree up to 6, and derivatives up to the fifth derivative.

Savitzky and Golay also pointed out that these convolutions can be performed using integer arithmetic, followed by just one division for each convolution.

In 1964, that was hugely important. On most machines of that time, floating point arithmetic was several orders of magnitude slower than integer arithmetic.

On modern computers, floating point arithmetic is just about as fast as integer arithmetic, so that advantage of Savitzky-Golay convolution has disappeared.

For small windows, the precomputed convolution coefficients should still make Savitzky-Golay convolution run faster than minimizing the sum of squared errors, especially if you're calculating only the smoothed time series and a couple of derivatives.

As the window size increases, Savitzky-Golay convolution takes longer. Using separate convolutions for each derivative also takes longer, but that implementation technique is optional even if the smoothing polynomial is calculated via Savitzky-Golay convolution.

Minimizing the sum of squared errors might well be faster for large windows, because the sums needed for minimization can be computed in constant time (independent of the window size) as rolling averages sums. In addition, for sufficiently large windows, differentiating and evaluating polynomials will be faster than using Savitzky-Golay convolutions to compute those derivatives.


[size=+1]Simple Implementation of Savitzky-Golay Smoothing/Differentiation[/size]

In 1964, the world's fastest computer was about ten thousand times as slow as the laptop I'm using today.

I implemented Savitzky-Golay smoothing/differentiation by minimizing the sum of squared errors and computing all non-trivial derivatives of the best-fit polynomial. My implementation might run slower than Savitzky-Golay convolution, but it produces exactly the same results (up to roundoff error). My code was able to take advantage of existing libraries for computing best-fit polynomials, and it shouldn't be hard to change my code to support more sophisticated methods that give less weight to errors farther away from the central point.

On their computer, Savitzky and Golay were able to process 10 points per second.

On my computer, with a very simple implementation that doesn't bother to compute the sums as rolling averages sums, I'm processing about 2000 points per second, even with 101 points per window. According to my profiler, almost all of my program's time is spent converting the floating point outputs to decimal and writing them to files.

femr2's best data for the NW corner of WTC7 consists of just over a thousand data points. That takes about half a second. According to what my profiler is telling me, using Savitzky-Golay convolution would save less than 200 milliseconds per run. How much more programming should I do to save my laptop from having to run for a few more seconds over the life of this program?

After testing my implementation of Savitzky-Golay smoothing/differentiation, I was ready to duplicate femr2's acceleration curve. Imagine my surprise when I was unable to do so.

Something was wrong. Was something wrong with my program, or was something wrong with femr2's description of the methods and parameters he used to derive his curve?

(Some of you may be way ahead of me.)

Tune in next week (or month), when we answer that question and examine some of femr2's more recent claims.

Eventually, we may see what more sophisticated methods might tell us, and whether any of this could possibly matter.
 
Last edited:
which of your graphs do you believe represents the early acceleration of the NW corner?

Would it be possible for you to show some result that you believe represents building acceleration of the NW corner?

From my point of view I'd like some information on NW corner acceleration.
All of my graphs show accelerations computed via Savitzky-Golay differentiation, together with femr2's alleged Savitzky-Golay curve. All of those curves, including femr2's, are representations of acceleration for the NW corner of WTC7. All of those curves are faulty. Before I could say which of those curves are best for your purposes, you'd have to explain your purposes.

To consider just one hypothetical example, suppose an honest researcher were to formulate technical questions whose resolution depended on analyzing accelerations in the neighborhood of collapse initiation at WTC7.

I would warn that hypothetical researcher against looking at Savitzky-Golay acceleration curves. As explained in my part 2, the wide windows made necessary by noise in the available data introduce artifacts in the neighborhood of sudden changes of acceleration. With the Savitzky-Golay accelerations I showed in my part 1, you'll see a less rapid increase in the downward acceleration after the collapse than was actually the case. This artifact comes from using Savitzky-Golay (or similar) differentiation with a wide window.

A more sophisticated algorithm might reduce or even eliminate those particular artifacts while introducing other artifacts or problems. I wouldn't be able to give sound advice until our hypothetical honest researchers explain their purposes.

Edited by LashL: 
Snipped quote of moderated content and response to same.
 
Last edited by a moderator:

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom