Trend Line Semilog Plot

[lluvia]

Hi there I'm new here and completely inexperienced--looked for a straightforward answer for this but couldn't find one..

I needed to graph a semilog plot that went on the y axis from 1-100 and on the x axis from 0-51,
then graph a few points and find a line of best fit. I did this on excel without any problem, using trendline for the line of best fit. I got a really nice graph that was exactly what I was looking for. I needed to get the data for the line of best fit so I copied the equation it gave me which was y=173.83*exp(-0.028x)

loggraph.jpg

Hopefully the image attached..but anyway when I used this equation to make a table of my data, using one column for y and one column with the x values from 0-51 & the answers are all wrong..for example
y(51)=41.68235135 but should be around 10, why is this happening? It gives me a great line of best fit but the equation doesn't match what the line is telling me..

[shg]

Welcome to the forum.

Posting pictures is no help unless you're looking for help with Photoshop. Take a few minutes to read How to get your questions answered quickly in the forum rules.

[MrShorty]

An actual spreadsheet is easier to debug than a picture.

What kind of chart are you using: scatter or line (http://office.microsoft.com/en-us/ex...478.aspx?CTT=1) I have seen many people come on here using a "line" chart from which it is often unclear what Excel used for "known x's" in performing the regression. When you don't know what Excel used for "known_x's", it becomes impossible to back out correlated values along the line. One of my first suggestions is to make sure you are using a scatter chart and not a line chart.

"Reading off" the data points from the picture and regressing them using the LINEST()/LOGEST() function, it almost looks like the chart trendline you report is using something different for "known_x" -- something that is not the given x values (almost looks like it is using 2x as the known_x's, but I can't be sure). My second suggestion, especially if this is something you need to do frequently, is to learn how to use the LINEST (and maybe LOGEST) functions to perform the regression directly in the spreadsheet. http://office.microsoft.com/en-us/ex...005209155.aspx