Graph Scale Problems: Solve 0.001-0.123 Range Issues

[_Mayday_]

Hey!

I am having real issues finding a good scale for a graph I am drawing. Most of the data is around 0.001 to 0.005 but then there are a few that are all way up at 0.123. I am really having trouble finding a scale for all of this. I can find a scale, but not without having to squash half of the graph together, which leads to a lot of inaccuracys in terms of plotting points. Does anyone have any suggestions?

 

[arunbg]

Perhaps you can avoid a few points that are too close together, and fix a suitable scale. Also, what shape are you graphing exactly? You could use graphing software, if accuracy is paramount.

 

[_Mayday_]

I have about 7 readings, 5 of them are near 0.001 and 2 are at the 0.123 end. I need to draw it out as well. It is a straigh line through the axis, that then levels off at the rop.

 

[arunbg]

Is there any way you can redo the experiment to get values nearby? You can have two graphs for the two separate regions if you want and show imaginary extrapolation, but it won't look nice.

 

[symbolipoint]

Maybe you should display your data for us, all of it, and somebody can give you more precise suggestions. Could your data fit well onto semi-log paper? Alternatively, could you treat one scale of values by finding logarithm of the x or the y first, and then graph onto cartesian paper?

 

[_Mayday_]

Okay, I have found an alright scale, it still looks a bit weird but ok. Now I need to find the gradient of this curve. It is extremely steep initially and then peaks and is extremely shallow at the end, how would I find the gradient of this? Would I need to draw a tangent? If so then where?

It kinda looks like this. (I only mean the shape not the topic the graph is on)

http://acs.confex.com/acs/werm05/techprogram/images/24298-0.gif

 

[Moonbear]

One way to handle issues with points that don't all fit on the same scale is to split your axis (and any lines connecting points where the axis splits, if appropriate) so that below the split you have a smaller scale, and above the split a larger scale. Alternatively, you can plot everything on a large scale to fit in the highest values, and then add an inset graph showing the lower values in detail on a smaller scale.