Thread: CAGR or other

Hi,

I have a question about compound annual growth rates:

Suppose I am talking about the number of articles published in Wikipedia annually between 2005 and 2010:

2005 = 10 articles
2006 = 5 articles
2007 = 20 articles
2008 = 21 articles
2009 = 19 articles
2010 = 20 articles

CAGR = (20/10)^(1/5)-1 = 0.149 = 15%

My question is, this obviously is not exponential growth, so is there a better measure?

Originally Posted by timc03

Hi,

I have a question about compound annual growth rates:

Suppose I am talking about the number of articles published in Wikipedia annually between 2005 and 2010:

2005 = 10 articles
2006 = 5 articles
2007 = 20 articles
2008 = 21 articles
2009 = 19 articles
2010 = 20 articles

CAGR = (20/10)^(1/5)-1 = 0.149 = 15%

My question is, this obviously is not exponential growth, so is there a better measure?

This is more than just a curve fitting problem. You need to apply some forecasting techniques and be able to forecast future values more accurately. A simple typical forecast analysis is attached:

Originally Posted by MaxJasper

This is more than just a curve fitting problem. You need to apply some forecasting techniques and be able to forecast future values more accurately. A simple typical forecast analysis is attached:

I suppose my question is a little simpler -I want to describe the average growth from 2005 to 2010, and I don't think CAGR is accurate - the growth is likely to be logistic over the long term, and not exponential, and as far as I understand it, CAGR describes exponential growth. So, is there a similar measure of average to describe logistic growth?

For logistic growth you need to specify upper and lower bounds!

Originally Posted by MaxJasper

For logistic growth you need to specify upper and lower bounds!

Can that take the form of either (from my above example)

2005 = 10 articles (lower) and 2010 = 20 articles (upper)

or

2006 = 5 articles (lower) and 2008 = 21 articles (upper)

I need to be able to describe the growth accurately in a summary statistic.

You can make all kinds of assumptions but in such cases assuming an upper and lower bound seems far from reality for any analysis or prediction or forecasting. However, if you can find a proper probability distribution function for your data then lower and upper bounds for logistic distribution will be 0 to 1.