To not use excel. So, I plotted this nifty data set:
Then, I asked excel to perform a simple linear regression and spit back the resultant coefficients. Then, using the proffered slope (-0.4257) and intercept (-177.7) I modeled my data, basically by taking my X values (depth) and plugging them into a linear equation. Which, surprisingly enough, didn't give me a linear plot:
Just a tad confused, I then asked excel to do a linear regression in the plot (something I quite frequently do) and again, spit out the regression coefficients. Notice the regression coefficients here (-37.04 for the slope and -138.38 for the intercept) are rather wildly different than the original coefficients excel provided.
Notice that although that line matches my (non-linear) model quite well, those coefficients produce wildly wrong estimates of my data when I actually crunch the numbers--not quite what I was looking for on a quiet Tuesday afternoon. I think I'll be sticking to matlab, thankyouverymuch.