yobit eobot.com

Announcement

Collapse
No announcement yet.

are you a math wizz, correspondence analysis

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Change the thermostat and go from there.

    Have you checked for any ECU codes?

    Comment


    • #17
      First of all yes, ALWAYS label your axis!

      I personally think you can draw some conclusions from the graph as there are groupings which do seem to make sense, but of course to be really professional you'd need to apply a statistical model to the data set.

      On the graph I see over by the frontal impact, you have numbers 4+5 (lower right and left extremities respectively - i.e. 'footwell damage from legs/knees flying forward!). The left impact all on it's own up the top with number 3 (pelvic girdle), presuming these are right hand drive cars makes sense, as a left impact would cause the body to bend over side ways into the passenger side causing damage to the lower spine (with a right hand impact the body would shunt right, onto the side wall and not bend so much initially). However with a right impact you'd likely see a head injury as it hits the side wall - your graph show this as '1' (skull) is grouped near the right impact triangle.

      Left and right upper extremities (2+6) are also grouped very close together and also close to the right impact plot (this makes sense). Likewise point 8 (vertebral column) is located in close relation to rear impact plot and furthest from all other points suggesting more vertebral injuries are caused by rear impacts - I.E. WHIPLASH!

      I can't think off hand the best correlation analysis model to use but I do know of an excellent book that will tell you:-

      HERE I know it says 'A Biologists Guide' but trust me, data is/are data! At uni we studied a model for animal population/lifecycle statistics by going out in town and logging all the car number plates we could! (i.e. each car has an age/lifespan/location made and colour etc) - was actually quite interesting. lol
      Last edited by Snorkel; 16 March 2008, 09:59.
      Live like ur gonna die, because ur gonna

      Comment


      • #18
        I should add - it would be interesting to add data for a front passenger (in a different colour) to this graph. I'd would expect '1' to be located nearer to 'left' than to 'right', also '3' should be closer to 'right' than 'left'.

        With rear passengers, you'd expect the lower leg injuries to become more central, with perhaps a slight drifting to 'left/right impacts' depending which side the passenger was on, and head injuries would perhaps close in on frontal impacts (again if there are no rear airbags) as it's usually the rear passenger who end up with their face hitting the two front seat head rests!

        Also I'd note whether the cars in the tests had passenger seat/rear airbags, as they'll have an effect on the head injury number data set (Though presumably the 'testers' would have kept it the same for all tests (if using test dummies)), but with the introduction on side airbags etc, it may be worth a paragraph or two in the discussion for noting improvements to the study and your expectations of future results from real crash data sets over the coming years.

        hope this was what you were after, but as I said if you need to show proof (personally I think the graph already does it with such distict groupings), then check out the book I mentioned.
        Live like ur gonna die, because ur gonna

        Comment


        • #19
          Also check your dipstick
          Rob

          Still working for the man!

          Comment


          • #20
            Ooooh, maths models...

            I once met someone who was doing trajectory modelling of RPGs and wanted to monitor / data gather at some live-fires I was at. She argued with real life for a while as it didn't match her model. Stupid girl.
            Cutting steps in the roof of the world

            Comment


            • #21
              Ah yes, there's nothing like the randomness of real life, I think it's called "chaos theory" ...........................but it's only a theory
              Rob

              Still working for the man!

              Comment


              • #22
                [quote=Apache;424824]Ooooh, maths models...

                I once met someone who was doing trajectory modelling of RPGs and wanted to monitor / data gather at some live-fires I was at. She argued with real life for a while as it didn't match her model. Stupid girl.[/quote]

                aint that wonderful when hat happens

                a youngish engineer seemed to think that a certain set of pumps WILL pump 16 l/s ,when they achieved 15 l/s the real life scenario was wrong as"matlab says it will work" goddam fool
                Non intercooled nothing.

                Comment


                • #23
                  Originally posted by BUSHWHACKER View Post
                  Change the thermostat and go from there.

                  Have you checked for any ECU codes?

                  Too young to die and too old to give a toss

                  Comment


                  • #24
                    Originally posted by Snorkel View Post
                    I should add - it would be interesting to add data for a front passenger (in a different colour) to this graph. I'd would expect '1' to be located nearer to 'left' than to 'right', also '3' should be closer to 'right' than 'left'.

                    With rear passengers, you'd expect the lower leg injuries to become more central, with perhaps a slight drifting to 'left/right impacts' depending which side the passenger was on, and head injuries would perhaps close in on frontal impacts (again if there are no rear airbags) as it's usually the rear passenger who end up with their face hitting the two front seat head rests!

                    Also I'd note whether the cars in the tests had passenger seat/rear airbags, as they'll have an effect on the head injury number data set (Though presumably the 'testers' would have kept it the same for all tests (if using test dummies)), but with the introduction on side airbags etc, it may be worth a paragraph or two in the discussion for noting improvements to the study and your expectations of future results from real crash data sets over the coming years.

                    hope this was what you were after, but as I said if you need to show proof (personally I think the graph already does it with such distict groupings), then check out the book I mentioned.
                    thank you so much, that was a great help. ive ordered the book its on its way. you a brilliant man.
                    Bring back firefly!!!
                    www.fireflyseason2.com

                    Comment


                    • #25
                      : Dunno! Sorry!!

                      Ferret
                      "Canis Auxilio Servimus"

                      Comment

                      Working...
                      X