Exponential Growth

[erin.carney1]

I have been using excel in order to determine the starting point of
exponential growth in data. For example, I have taken this data,
determined the log value. I used the log value with the time to
determine the slope and r^2 value. Then used the r^2 value to detect
when the data demonstrated the beginning of the exponential phse. The
only problem is the results aren't incredibly accurate. I was wondering
if anyone else had an idea about how i could detect when this data
starts to grow exponentially. (like LOGEST, or transformed regression
model)

Thanks for any help

 

[joeu2004]

I have been using excel in order to determine the starting point of
exponential growth in data. For example, I have taken this data,
determined the log value. I used the log value with the time to
determine the slope and r^2 value. Then used the r^2 value to detect
when the data demonstrated the beginning of the exponential phse. The
only problem is the results aren't incredibly accurate.

What do you mean by "not accurate"?

I was wondering
if anyone else had an idea about how i could detect when this data
starts to grow exponentially. (like LOGEST, or transformed regression
model)

That is more or less what I have done. I start by graphing the data to
eyeball the starting point. Then depending on the situation, I might
use Excel's graphical trendline feature, or I might use LOGEST and RSQ.
No need to convert to log values.

FYI, unless theory tells me to expect exponential growth, in some
cases, an examination of the graph has led me to discover that some
"exponential" data is best fit by multiple linear lines. That has led
me to new theories for why that might be the case. My point is: start
with the graphical representation and be open to new interpretations of
the data.