First Derivative

**sramalin** · 06-07-2011, 04:31 AM

Hi,

I have two columns of data (P, t) and am trying to get the derivative dP/dt defined (P3-P1)/(t3-t1), that is skipping points inbetween to achieve two objectives, one to reduce the amount of data i have and two to achieve smoothing of my dP/dt vs t curve. I assume writing a VB code would be easy to do it, but I dont have any experience doing the same. Could someone help me?

Performing derivative as dP/dt = (P2-P1)/(t2-t1) was easy. I basically used a macro to insert a new row after every data point (copied from this forum). Then just setup a formula in this new row of the subsequent column to do (P2-P1)/(t2-t1). I have attached a file showing how I did it. But doing the same when I want P3-P1)/(t3-t1) is proving a big pain!!!

Someone please help me..

Thanks
Karthik

P.S: Don't bother about the mismatch in data, I just wanted to show how I did it..

**shg** · 06-07-2011, 08:08 AM

Welcome to the forum.

Why not it D5 and copy down, =(B6-B4)/(C6-C4)

**sramalin** · 06-07-2011, 09:55 AM

Originally Posted by shg

Welcome to the forum.

Why not it D5 and copy down, =(B6-B4)/(C6-C4)

Because that doesnt reduce the amount of data I have, which I want to accomplish also apart from calculating the derivative..
so D5= (B6-B4)/(C6-C4) then the subsequent differential would be (B9-B7)/(C9-C7), (B12-B10)/(C12-C10) and so on so forth. This would reduce my data as well..

Thanks for your help.

**shg** · 06-07-2011, 10:07 AM

I'm missing the objective; why not reduce the data first, then take the derivative of what's left?

You could reduce the data by decimation, or by taking a centered moving average of every 3, 5, whatever, points.

**sramalin** · 06-07-2011, 01:43 PM

Originally Posted by shg

I'm missing the objective; why not reduce the data first, then take the derivative of what's left?

You could reduce the data by decimation, or by taking a centered moving average of every 3, 5, whatever, points.

Yes I have already done that. The data I have is after I did 3 point moving average. I am not performing 4 or 5 point moving average because my analysis involves looking at the exact position of slope change in the data and after trials on various data, 3 point averaging works best. But the amount of data is still large enough (and hence more scatter) that I am not able to recognize the exact position of the slope change when I plot the derivative as a function of the variable. So if I do a derivative by skipping a point and then plot the derivative as the function of the variable, I reduce the data further and I have my bosses quoting that it smoothens the plot as well. And hence this painful process.

Thanks for your help. Let me know