I have data looks like uploaded image (scatter plot).
In scatter plot, I can clearly distinguish two different Gaussian curve, but origin or excel cannot.
Data is sorted by x axis value, so cannot simply separated by cut bottom or top part of data.
Peak fitting function in origin also not helpful, because it consider this data as noisy single Gaussian curve, not two different curve merged together.
I am looking for software or tool that separate this data to two data, like human can do.
Is there any automated methods not human brute force, like select data points one by one myself?
thanks.
Jae Hwan Jeong
Click the cell you wish to split in the table. Navigate to the Layout tab. Click Split Cells in the Merge group. Select the number of columns and rows you want in the Split Cells dialog, then click OK.
On Excel, perform the following steps:
1. Navigate to Home > Find & Select > Find.
2. Select Format from the drop-down menu.
3. Select Alignment > Merge cells > OK from the drop-down menu.
4. To get a list of all merged cells in your worksheet, click Find All. Excel picks the combined cell in your worksheet when you click an item in the list. You can now separate the cells.
Jae Hwan Jeong, you can separate your data by introducing two Gaussian functions according to approximate parameters from your graph (peak amplitude, x shift, and width values), drawing them with your given data. You can find good correspondence between the first and second curves by adjusting two sets of fit parameters. After finding the Gaussian fitting parameters, you can calculate the difference between calculated (fitted) and given values and make data separation based on the difference values.
If you have questions, I'm ready to answer.
Hello Jae Hwan Jeong .
I am confused by something you said: "Data is sorted by x axis value, so cannot simply separated by cut bottom or top part of data." Does this mean that the data was originally in a table format such as:
x y
50 60
51 90
or
x y
50 60
50 90
What I am asking is. How was the data generated? Perhaps going back to how the data was generated or captured is the solution, instead of trying to deal with the data after the table is created. I am also confused by what kind of measurement process can give rise to a superposition of two Gaussian-like curves. Could the measurement process be defective, i.e., the gain of the instrument is changing or something like that?
Regards,
Tom Cuff
Thanks for your answers.
Valeriy Sterligov This method looks still require some manual process, but thank you. I'll try that If I cannot find more automated way.
Thomas Cuff sorted by x axis value mean, for example,
Current I have data table like this :
x y
5 0
10 1
12 1
15 2
17 5
20 5
22 15
27 10
...
And I want separate data to two different table - looks like this :
x y
10 1
15 2
20 5
27 10
...
and
x y
5 0
12 1
17 5
22 15
...
Assume that numbers in above tables are following two different gaussian curve.
This data is height distribution histogram graph from AFM image (x:height, Y:pixel count).
My sample has grain, slightly higher than surrounding flat surface.
I know many SPM image processing software (such as gwyddion) has grain tool that distinguish each grains and make statistics for each grain, but as you can saw in graph, height distribution (noise) for each grain (and base surface) is too large to software cleanly separate each other, (it is possible, but require many manual tweaking) even grain looks obvious for human eye.
Jae Hwan Jeong
>This method looks still require some manual process, but thank you. I'll try that If I cannot find more automated way.
Such numerical parameters of two Gaussians will be helpful even more than raw AFM data. You can use them for surface characterization. And after such kind of filtering that I propose to you, you can perform software Gaussian fit of filtered data, so the precision of final Gaussian parameters will be higher.
This software can do what you need! Example included.
www.lerenisplezant.be/fitting.htm
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials
- Semiconductor