# Finding correlation, triggers and patterns between two (or more datasets)

I am trying to find correlations between multiple data sets, for this example i am using two pieces of data, the closing stock price for Coca-Cola and Pepsi for the last year.

I've been pointed in the direction of using the =CORREL() function in excel which has given me an indication that there is a correlation as it returns the value 0.869783523 but does not give me any clue as to triggers or patterns in the data which is what i am actually interested in.

Is there function or a workflow that i can implement in excel to give me a more detailed read out giving me an idea of triggers and/or patterns?

• Not sure I understand the question, but if you need more stats output then add Data Analysis Toolkit to Excel. – zx8754 Sep 16 '13 at 15:29
• @zx8754 - In a nutshell im trying to find if there are any triggers and / or patterns in the data, ie data A does this data B tends to do this. Is there a function or workflow in excel to analyze help this ? - Thanks for the Data analysis toolkit, ill check that out now – sam Sep 16 '13 at 15:32
• I think what you need is Display the trendline equation on the chart – zx8754 Sep 16 '13 at 15:42
• I think you need to better define what you mean by triggers/patterns. There are a million statistical analyses you could do so telling us exactly what you are after would be much better. – user25658 Sep 16 '13 at 15:52
• @BabakP - Im not particularly well versed in statistics, but ill give articulating it a try - In the simplest terms in trying to find if there are reliable or predictable instances where dataA moves and it triggers a move in dataB – sam Sep 16 '13 at 15:56

From your question and further comments, you are not looking for classic correlation, you are looking for some kind of causal, or at least temporal correlation. At first, the data you have seems to be just a vector of two variables; Coke and Pepsi closing stock prices. Looking at the Pearson product-moment correlation coefficient , which is what CORREL calculates, will give you the linear relationship of one variable with the other, but no more.