Smoothing curves of raw data (no trendline)

**J0hnnyV1ntage** · 12-07-2012, 10:21 AM

Hello everyone!

I need your expertise on how to smoothen out my data curves. I'm doing my master thesis this year and time is short, and the graphs I have to smoothen out are sometimes 10000 minutes long with intervals of 1/3 of a minute

In order to show my temporary work, the graphs have to be readable, but can't be fitted since they want to see the actual data. In the file you see an example of this data with large peaks from time to time. The graph doesn't have to be perfectly smooth, in fact it's peferred if it's not, but the large peaks should be decreased to a minimum since they're only data points caused by shifts (what I'm trying to say is the graphs should still be credible to a certain degree).

Can anyone help me?

Thanks

J0hnnyV1ntage

Curve_With_Peaks.xlsx

**Steve N.** · 12-09-2012, 07:10 PM

Johnny,

I often use a simple moving average to smooth peaks/outliers in my data. I hope this is of some help...

Here's a sample of your data (I chopped off about the last 1,500 lines to make the file smaller) re-plotted to a 50 event simple moving average. I added columns D, Q, and U to hold the moving average of each column in your original plot. I then started the formula at row 135 of each new column to average the first 50 events, for example the formula in D135 =AVERAGE(C86:C135) is averaging the values in that 50 cell (event) range. Copied down, this averages the next set of 50 events from C87:C136, C88:C137, and so on. Same applies to columns Q and U. You could even make it a 100 event moving average - pick a number/starting point that suits your needs...

I did not count C85 as an event since the first event from C85 starts in C86.

Oh, and in your original attachment, it seems the Graph legend on the chart does not actually match the column headers in the data. Graph 1 in the legend is tied to column 'O' titled "Graph2". In the attached s/s I left the columns as you had them labeled and changed the chart.

Is this something you can use?

Steve

**J0hnnyV1ntage** · 12-10-2012, 10:00 AM

Thanks Steve!

I actually didn't think it could be that simple, but the result looks great

and you're right about the original, I had to censure my data and I had to do it rather quickly, so I accidently switched them. For the points at the beginning I guess I could just copy them since the noise is rather low there.

Many thanks again for your quick response

J0hnny