Converting axis tick mark labels from natural log to original values

[jeffrey hessParticipant]

I prepared quite a few graphs of chemical concentrations of aromatic compounds in oil based on natural log values for the measured constituents versus time, then applying a regression to each trend to evaluate their dissolution rates from oil. I have received several comments that the graphs are incorrect as dissolution is not linear! The above approach is recommended by EPA to determine decay rates for first order decay, which is fundamentally the same.  I believe they are not reading the label to the Y axis to understand these are log values (it is also in the figure title to the graphs).

So my thought was to display the y axis as the reported values prior to taking the log of the values. I am stumped on how to accomplish this, if it is even possible. I tried using a second Y axis but I could not work out how to get this to work.

Suggestions?

[dgteamModerator]

Perhaps this will help …

We have an example file that illustrates using the exponential fit function. This is essentially using a first-order decay equation, which we used in his file to estimate a half-life.

 

In one graph, the fit is displayed on the linear scale;

 

in the other, it is displayed on the log scale (which gives a straight line).

The Exponential Fit option in the Fit command linearized the data behind the scenes, but you can display it in either format by changing the ‘X type’ in the Axis settings.

If this is still not quite what you’re looking for, it might be helpful to post an image so we can see the graph.

How to post an image

 

 

[jeffrey hessParticipant]

I was trying to show both scales on a single figure. I did finally get it to work by adding another Y axis and then hiding everything (once the points and regression aligned) on the new axis except the axis and scale. The resulting graph is this:

[Author]

Posts