Introducing the CUSUM chart

The Cumulative Sum chart (CUSUM) is a very useful (and powerful) tool to pick up small changes in trends of noisy data- and it works just as well if there are large trend changes. Despite what some might say, it is actually relatively easy to apply this technique in a spreadsheet to analyse your data, although one must understand how to interpret the graph.

The CUSUM is very useful to identify important dates or times (turning points) where a trend change occurred, especially where one finds that this is not obvious when looking at the time series data.
You can use it to guage the likelihood that other noisy variables could be causes if one can pick up similar trend changes in these variables. It finds applications in data analysis especially of process data and is a tool that can be used by Six Sigma practitioners.

In my next post I will elaborate on how to implement this useful technique in a spreadsheet such as Excel or OpenOffice, and also show an example, and more importantly how to interpret the graph, so stay tuned. For now, here are some related links.

Cusum Link 1 (NIST)
Cusum Link 2 (iSixSigma)