Smoothing or filtering data for graphs
Solved
Luccie
Posted messages
16
Registration date
Status
Membre
-
Psykokopath -
Psykokopath -
Hello,
So I apologize in advance if I'm not in the right forum :-)
So here's my issue: I have very heterogeneous data, for example my graph looks like this: oops, I can't insert an image, so I'll share some data instead, it's better than nothing: (I'm sharing quite a bit because it really helps to understand in a graph)
4 17.720045
4.04 17.102631
4.08 15.787653
4.12 18.681542
4.16 17.755281
4.2 16.763055
4.24 15.524175
4.28 15.297059
4.32 15.297059
4.36 14.807093
4.4 15.890249
4.44 14.56022
4.48 15.402922
4.52 14.142136
4.56 15.787653
4.6 14.534442
4.64 13.5
4.68 14.5
4.72 14.008926
4.76 13.009612
4.8 14.008926
4.84 14.142136
4.88 16.007811
4.92 17.029386
4.96 17.18284
5 16.007811
5.04 16.124515
5.08 16.03122
5.12 16.77051
5.16 16.568042
5.2 16.62077
5.24 18.11077
5.28 17.029386
5.32 17.67767
5.36 17.67767
5.4 17.117243
5.44 16.492423
5.48 17.18284
5.52 15
5.56 15.297059
5.6 13.72953
5.64 14.637281
5.68 14.080128
5.72 14.5
5.76 14.317821
5.8 13.536986
5.84 11.672618
5.88 14.534442
5.92 14.008926
5.96 12.658989
6 13.009612
6.04 14.5
6.08 13.536986
6.12 14.035669
6.16 13
6.2 14.142136
6.24 13.647344
6.28 14.534442
6.32 15.132746
6.36 15.811388
6.4 13.647344
6.44 14.637281
6.48 13.341664
6.52 13.341664
6.56 15.402922
6.6 14.56022
6.64 15.041609
6.68 12.041595
6.72 14.56022
6.76 13.341664
6.8 14.142136
6.84 14.637281
6.88 13
6.92 15.074813
6.96 15.074813
7 17.117243
7.04 15.532225
7.08 16.688319
7.12 16.124515
7.16 15.008331
7.2 16.62077
7.24 15.5
And actually, you can see the trends (up or down) but then the close variations (peaks) are actually due to the sensitivity of the sensors and noise. So I would like to obtain a more uniform curve to analyze the results... And I'm struggling, I've tried with Excel and Scilab to no avail. I've read some stuff about the least squares method but I don't understand anything at all and since I'm not sure whether we can obtain a curve in the end and not just a straight line, I'm hesitant to spend hours trying to understand (given that I've been looking for a solution everywhere for two days, I want to avoid false leads)...
I hope I've been clear, thank you so much if you have an idea...
So I apologize in advance if I'm not in the right forum :-)
So here's my issue: I have very heterogeneous data, for example my graph looks like this: oops, I can't insert an image, so I'll share some data instead, it's better than nothing: (I'm sharing quite a bit because it really helps to understand in a graph)
4 17.720045
4.04 17.102631
4.08 15.787653
4.12 18.681542
4.16 17.755281
4.2 16.763055
4.24 15.524175
4.28 15.297059
4.32 15.297059
4.36 14.807093
4.4 15.890249
4.44 14.56022
4.48 15.402922
4.52 14.142136
4.56 15.787653
4.6 14.534442
4.64 13.5
4.68 14.5
4.72 14.008926
4.76 13.009612
4.8 14.008926
4.84 14.142136
4.88 16.007811
4.92 17.029386
4.96 17.18284
5 16.007811
5.04 16.124515
5.08 16.03122
5.12 16.77051
5.16 16.568042
5.2 16.62077
5.24 18.11077
5.28 17.029386
5.32 17.67767
5.36 17.67767
5.4 17.117243
5.44 16.492423
5.48 17.18284
5.52 15
5.56 15.297059
5.6 13.72953
5.64 14.637281
5.68 14.080128
5.72 14.5
5.76 14.317821
5.8 13.536986
5.84 11.672618
5.88 14.534442
5.92 14.008926
5.96 12.658989
6 13.009612
6.04 14.5
6.08 13.536986
6.12 14.035669
6.16 13
6.2 14.142136
6.24 13.647344
6.28 14.534442
6.32 15.132746
6.36 15.811388
6.4 13.647344
6.44 14.637281
6.48 13.341664
6.52 13.341664
6.56 15.402922
6.6 14.56022
6.64 15.041609
6.68 12.041595
6.72 14.56022
6.76 13.341664
6.8 14.142136
6.84 14.637281
6.88 13
6.92 15.074813
6.96 15.074813
7 17.117243
7.04 15.532225
7.08 16.688319
7.12 16.124515
7.16 15.008331
7.2 16.62077
7.24 15.5
And actually, you can see the trends (up or down) but then the close variations (peaks) are actually due to the sensitivity of the sensors and noise. So I would like to obtain a more uniform curve to analyze the results... And I'm struggling, I've tried with Excel and Scilab to no avail. I've read some stuff about the least squares method but I don't understand anything at all and since I'm not sure whether we can obtain a curve in the end and not just a straight line, I'm hesitant to spend hours trying to understand (given that I've been looking for a solution everywhere for two days, I want to avoid false leads)...
I hope I've been clear, thank you so much if you have an idea...
Configuration: Windows XP Firefox 2.0.0.16
5 réponses
The problem is that I don't have the 2000 version! But I can specify this for the 2003 version:
1) when using the polynomial trendline, you can choose a value between:
* order 2, which is the default and indeed gives a kind of straight line,
* order 6, the highest, which gives a nice wave.
2) when using the moving average trendline, you can choose a value starting from order 2, which is the default and closely follows the initial curve; then the higher the order, the more the curve flattens and starts at a higher abscissa.
3) attached is my file https://www.cjoint.com/?ihqTucYTMT on which I worked to provide you with these explanations, and whose graph includes:
* the initial curve (with Data Series Format/Patterns/Smoothing) in thin blue line,
* the polynomial trendline of order 6, thick red line, for reference,
* the moving average trendline of order 2, thick black line, which is still quite discontinuous,
* the moving average trendline of order 3, medium green line, which seems to me the most interesting for you,
* the moving average trendline of order 60, medium pink line, which goes from abscissa 60 to abscissa 82.
Do you have the equivalent in Excel 2000? But for the work you do, you should seriously consider upgrading your version of Microsoft Office!
Warm regards,
--
Retirement is great! Especially in the Antilles... :-)
Raymond
1) when using the polynomial trendline, you can choose a value between:
* order 2, which is the default and indeed gives a kind of straight line,
* order 6, the highest, which gives a nice wave.
2) when using the moving average trendline, you can choose a value starting from order 2, which is the default and closely follows the initial curve; then the higher the order, the more the curve flattens and starts at a higher abscissa.
3) attached is my file https://www.cjoint.com/?ihqTucYTMT on which I worked to provide you with these explanations, and whose graph includes:
* the initial curve (with Data Series Format/Patterns/Smoothing) in thin blue line,
* the polynomial trendline of order 6, thick red line, for reference,
* the moving average trendline of order 2, thick black line, which is still quite discontinuous,
* the moving average trendline of order 3, medium green line, which seems to me the most interesting for you,
* the moving average trendline of order 60, medium pink line, which goes from abscissa 60 to abscissa 82.
Do you have the equivalent in Excel 2000? But for the work you do, you should seriously consider upgrading your version of Microsoft Office!
Warm regards,
--
Retirement is great! Especially in the Antilles... :-)
Raymond
I couldn't open your file; it says it no longer exists, but I applied what you told me and I believe, it seems to me, that it looks very good... But I’d rather confirm it when I’m calm; I don't want to celebrate victory after so many hours of headache ;-)
No kidding, it looks great. I still need to think about which one is the most suitable for my study, and that won't be easy, but anyway, really a big thank you, especially for your explanations, with which I will be spending several hours one-on-one deciding on the right trend line and the right order (I'll finally understand what order is for).
And about the version, yes, I know I actually keep switching computers, and at this moment I was on this one, but I have 2 different versions on the computers in front of me, and I have more different ones at work, so I really don't pay attention to the version I'm using and I fiddle with it every time :-)
Thanks again, you’ve greatly brightened my day all of a sudden, and I’m even going to allow myself a break for the evening ;-)
Have a good evening!