Page 1 of 4 123 ... LastLast
Results 1 to 15 of 49

Thread: An example of how digitized data can be filtered for presentation, both good and bad....

  1. #1
    Join Date
    Jan 2012
    Posts
    4,096

    An example of how digitized data can be filtered for presentation, both good and bad....

    One of the issues we've been working through since completion of the Research-a-thon (with the assistance of the data analysis experts at Xcitex, creators of ProAnalyst) is how to filter the derived data for presentation. Here is a simple example.

    Below is one of Denny Albert's swings, a very sold strike that landed just 13 feet left of the target line (according to Trackman anyhow!). I stopped the video just after initial contact (you can see the ball compressing). The ProAnalyst software was asked to "track" the spot on the toe of Denny's driver (in red). Measuring in pixels, ProAnalyst gathers the horizontal and vertical location of the center of the spot for each frame. Based on this raw data, the ProAnalyst software can determine the spot's path and speed.

    In the graph below, the y-axis represents the toe spot's horizontal location in the frame, measured in pixels. You can see the graph line flatten a bit after impact, indicating that the clubhead has slowed from the collision with the ball. The data graphed here looks nice and smooth, right?

    Name:  raw data.PNG
Views: 846
Size:  123.2 KB

    However, when the horizontal velocity of the toe spot is calculated, the graph of that calculated data is very jumpy. That is caused by quantization noise, small errors in the toe spot's location frame-to-frame that occur in the data gathering process. All systems that gather digitized data (a succession of "snapshots" linked together) from real life analog data (that flows in a continuous line) experience quantization noise. To recreate more "real life"-looking, "smooth" data, data filtering is required.

    Name:  raw velocity.PNG
Views: 912
Size:  42.4 KB


    However, how the data is smoothed is of utmost importance to accurately represent reality. Inept filtering can greatly distort reality. Here is an example of "bad filtering" for this set of a data. A mean-median filter has been applied that averages the calculated velocity over several data points (before and after) to smooth out the jumps. See how nice it looks? But, was the change in velocity anywhere close to that gradual? Of course not: the change in velocity was much more abrupt.

    Name:  mean median.PNG
Views: 737
Size:  37.4 KB


    A more appropriate filter for this data set is something called a "low pass" filter. I applied a low pass filter to the same data and got this result; not quite as smooth and pretty as the mean-median filter, but MUCH more representative of reality.

    Name:  butterworth.PNG
Views: 790
Size:  43.4 KB


    Now, what kind of filter does it look like AMM3D used to create the smooth, nicely rounded curve of pelvic rotational velocity in the graph below (y-axis is degrees per second, x-axis is time) from pelvis angle data that abruptly changed before the top of the swing and after impact but not DURING the downswing? hmmmm...

    Name:  amm 3d rotational.PNG
Views: 889
Size:  468.1 KB


    Y-axis is angle of rotation in degrees, x-axis is time:

    Name:  amm 3d hip angle.PNG
Views: 760
Size:  360.1 KB








    Jeff

  2. #2
    Thats interesting, and part of the reason I am apprehensive of drawing conclusions at this point in time.

  3. #3
    This is something we in electronics deal with all the time. Many people thought the quality of CD's for audio was so good, but because of the quantization noise, and sampling theory, you are pointing out, many audio people think CD's are junk. A true audio person gets even more upset if you try to have them use a mp3 because the sampling and compression is even worse.

    You should consider adding an analog audio engineer to look at your results. Someone who is well versed in sampling theory and quantization noise. They may help with the results.

  4. #4
    Lloyd Higley Guest
    Filtering will not smooth out the digital wave form, what you need to do is to increase the sampling rate or frequency

  5. #5
    Join Date
    Jan 2012
    Posts
    4,096
    I guess Cheetham claims to not filter his rotational velocity data, but Mike Jacobs posted yesterday afternoon why the rotational velocity graph is inconsistent with the pelvis angles graph: the data is different. For some reason, the AMM3D pelvic rotational velocity graph is not derived from the data in the pelvis angles graph. Very sneaky: AMM3D creates its own "apples and oranges" but doesn't tell anybody! In that case, it doesn't matter what filter you use (or don't use).

    Anyhow, maybe Cheetham should filter his data, because the two graphs on the right show the club decelerating before impact, which only happens if the club hits the ground before the ball. Sorry, Phil, what your graphs show doesn't happen in the real world, unless you analyze muffed shots!

    Name:  decel before impact.PNG
Views: 1886
Size:  129.0 KB


    This one is even better, purportedly an "analysis" of Jim McLean's 6 iron swing. What the hell is wrong with Jim McLean if his club decels that much before impact??? Uh, maybe it's the machine. Also, how does McLean's club decelerate then re-accelerate just before impact (see the little uptick in the purple line? does it with his hand, too!)???? That just can't happen, yet this shit is published and fools like Manzella gush all over it.


    Name:  mclean.PNG
Views: 1809
Size:  512.8 KB





    Jeff

  6. #6
    Maybe McLean did a Charles Barkley on that one . Seriously you are talking about guys in the golf industry that maybe scuffed a pass at high school.

  7. #7
    Join Date
    Aug 2012
    Location
    Pacific Palisades, California
    Posts
    73
    Quote Originally Posted by Jeff Martin View Post
    I guess Cheetham claims to not filter his rotational velocity data, but Mike Jacobs posted yesterday afternoon why the rotational velocity graph is inconsistent with the pelvic rotation angle graph: the data is different. For some reason, the AMM3D pelvic rotational velocity graph is not derived from the data in the pelvic rotation angle graph. Very sneaky: AMM3D creates its own "apples and oranges" but doesn't tell anybody! In that case, it doesn't matter what filter you use (or don't use).

    Anyhow, maybe Cheetham should filter his data, because the two graphs on the right show the club decelerating before impact, which only happens if the club hits the ground before the ball. Sorry, Phil, what your graphs show doesn't happen in the real world, unless you analyze muffed shots!

    Name:  decel before impact.PNG
Views: 1886
Size:  129.0 KB


    This one is even better, purportedly an "analysis" of Jim McLean's 6 iron swing. What the hell is wrong with Jim McLean if his club decels that much before impact??? Uh, maybe it's the machine. Also, how does McLean's club decelerate then re-accelerate just before impact (see the little uptick in the purple line? does it with his hand, too!)???? That just can't happen, yet this shit is published and fools like Manzella gush all over it.


    Name:  mclean.PNG
Views: 1809
Size:  512.8 KB





    Jeff
    Dear Jeff,

    All the above discussion and data continue to provide me a great extension to the passion seeking 'golf truth', and I sincerely am thankful and look forward to it continuing.

    BUT, is there some way to 'dumb down' the profanity of you and your followers as I show my wife and family how much golf science is being studied by so many learned people NOW???

    Finally in keeping with my personal goal of always ADDING VALUE that I believe is already scientifically proven, I suggest/implore you to engage what you have called one of your 'geniuses', Kelvin to weigh in on the NATURAL FREQUENCIES, anatomically of the body parts and groups under this extreme scrutiny.

    IMO, but an educated one, you, Kelvin and all your 'profane', oh pardon my slip, 'profound' followers are about to find that these frequencies are VERY low, and well within the capability to be measured and displayed as 'golf truth'.

    Looking very forward to your 'teams' responses and comments about the 'dynamics' and frequencies ESPECIALLY of the hips from the experts in anatomy, including the effects/impacts in understanding the global and local reference systems.

    Sincerely, appreciatively and without profanity,
    Art

  8. #8
    Join Date
    Jan 2012
    Posts
    4,096
    art-

    As my kids were growing up, our family greatly enjoyed watching together "South Park" and "The Sopranos". Need I say more?


    Jeff

  9. #9
    Join Date
    Aug 2012
    Location
    Pacific Palisades, California
    Posts
    73
    Quote Originally Posted by Jeff Martin View Post
    art-

    As my kids were growing up, our family greatly enjoyed watching together "South Park" and "The Sopranos". Need I say more?


    Jeff
    Dear Jeff,

    Certainly explains YOUR habits, but IMO, the golfing community, beginner to professional seems to communicate differently. And as I am sure you realize, unless there is an 'impedance match', not much energy (learning) can be flowing.

    But much more importantly to those we are BOTH trying to reach, PLEASE don't quit with your limited response, and sincerely and appropriately address the bodies 'dynamic' and natural frequency characteristics for the benefit of ALL of us.

    I believe you and your experts (maybe not scientists) there is a wide range of REQUIRED frequency response requirements when trying to measure 'hip rotation' and club/ball impact.

    Finally, please remember, my invitation for you, Kelvin and whomever to meet here in Los Angeles near the airport 'is rain check. protected' .

    Off to the doc, and some afternoon golf research.

    Sincerely,

    Art

  10. #10
    nmgolfer Guest
    Jeff,

    Excellent post. Unfortunately it is probably well over the head of the person who most needs to understand it. Have you looked at modal decomposition as a means of filtering your digital signals> EMD or SSA? Since there is temporal consistency of the samples should be a piece of cake. Nice to see the latest sadlowski confirming your conclusion. The bmanz should just shuddup because every time he opens his pie hole he further embarrasses himself. Keep up the good work!

  11. #11
    Join Date
    Jan 2012
    Posts
    4,096
    Quote Originally Posted by art View Post
    Dear Jeff,

    Certainly explains YOUR habits, but IMO, the golfing community, beginner to professional seems to communicate differently. And as I am sure you realize, unless there is an 'impedance match', not much energy (learning) can be flowing.

    But much more importantly to those we are BOTH trying to reach, PLEASE don't quit with your limited response, and sincerely and appropriately address the bodies 'dynamic' and natural frequency characteristics for the benefit of ALL of us.

    I believe you and your experts (maybe not scientists) there is a wide range of REQUIRED frequency response requirements when trying to measure 'hip rotation' and club/ball impact.

    Finally, please remember, my invitation for you, Kelvin and whomever to meet here in Los Angeles near the airport 'is rain check. protected' .

    Off to the doc, and some afternoon golf research.

    Sincerely,

    Art

    art-


    Sorry, but I don't give a fuck about the "golfing community" and how it "communicates".



    Jeff

  12. #12
    nmgolfer Guest
    I don't think you mean frequency response do you Art? He's not analyzing a vibration nor does the natural frequency of (anything) pertain to same.

    Quote Originally Posted by art View Post
    Dear Jeff,

    Certainly explains YOUR habits, but IMO, the golfing community, beginner to professional seems to communicate differently. And as I am sure you realize, unless there is an 'impedance match', not much energy (learning) can be flowing.

    But much more importantly to those we are BOTH trying to reach, PLEASE don't quit with your limited response, and sincerely and appropriately address the bodies 'dynamic' and natural frequency characteristics for the benefit of ALL of us.

    I believe you and your experts (maybe not scientists) there is a wide range of REQUIRED frequency response requirements when trying to measure 'hip rotation' and club/ball impact.

    Finally, please remember, my invitation for you, Kelvin and whomever to meet here in Los Angeles near the airport 'is rain check. protected' .

    Off to the doc, and some afternoon golf research.

    Sincerely,

    Art

  13. #13
    Jeff,

    In the 3 graph sequence, I believe Nate is the one who floated that over at Manzella's, I asked where in the swing do the hips start to decelerate, if the graph of top to impact is linear to the postion of the body it would be at P6, from there on the hips are slowing.

    I don't believe that at all.

  14. #14
    Join Date
    Jan 2012
    Posts
    4,096
    Quote Originally Posted by gmbtempe View Post
    Jeff,

    In the 3 graph sequence, I believe Nate is the one who floated that over at Manzella's, I asked where in the swing do the hips start to decelerate, if the graph of top to impact is linear to the postion of the body it would be at P6, from there on the hips are slowing.

    I don't believe that at all.


    I think the AMM3D "Kinetic Sequence" graphs are worse than worthless: they are outright misleading. The AMM3D "angles" graphs seem to tell a much more accurate, and VERY DIFFERENT, story, yet they've been kept hidden, or at least WAY in the background. Are the kinetic sequence guys being deceptive or are they just clueless??? Or both?



    Jeff

  15. #15
    Join Date
    Aug 2012
    Location
    Pacific Palisades, California
    Posts
    73
    Quote Originally Posted by Jeff Martin View Post
    art-


    Sorry, but I don't give a fuck about the "golfing community" and how it "communicates".



    Jeff
    Dear Jeff,

    Peace,

    Sincerely,
    Art

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •