Alteryx Designer Desktop Discussions

Find answers, ask questions, and share expertise about Alteryx Designer Desktop and Intelligence Suite.
SOLVED

Strongest Correlation values on output of pearson correlation tool

poojasingh111
6 - Meteoroid

I'm learning Data investigation tools and completely new to it 

how to determine strong correlation values between two variables(A-B),(A-C) based on the output below and why

 

poojasingh111_0-1655388527237.png

 

7 REPLIES 7
Ladarthure
14 - Magnetar
14 - Magnetar

Hi @poojasingh111,

 

correlation is a measure to see if there is a link between values. It goes from -1 to +1 the closer it is to 1 the bigger the correlation is.

 

You can read more about it here :

 

https://en.wikipedia.org/wiki/Correlation

 

https://help.alteryx.com/20221/designer/pearson-correlation-tool

 

IraWatt
17 - Castor
17 - Castor

Hey @poojasingh111,

The closer to 1/-1 the higher the positive or negative correlation:

IraWatt_0-1655388773260.png

You have 1's in the dataset as everything correlates directly with itself.

DataNath
17 - Castor

@poojasingh111 the closer the value is to 1 (positive correlation) or -1 (negative correlation), the stronger the correlation is. This is because 1 represents 2 linked variables I.e. the change of one represents a directly proportional change in the other. You should always get a diagonal of 1s through the middle as a variable will obviously always be perfectly correlated with itself.

poojasingh111
6 - Meteoroid

still I didn't understand, can anyone help me to understand through given example in the post, if I take two variables(A-B), are the strongest values calculated as vertically and horizontally

DataNath
17 - Castor

@poojasingh111 in this example, the 2 most correlated variables are A-C as the value is the closest to 1 or -1 (-0.897...). As their value is negative, this means they have a negative correlation i.e. as one increases, the other decreases.

poojasingh111
6 - Meteoroid
poojasingh111_1-1655395419971.png

 

DataNath, are you taking black penciled(vertical) or Red penciled(horizontally to calculate the closest value?If I understood it clearly any value near to 0 will be treated as less strongest and any value near to -1 or 1 is most strongest?

DataNath
17 - Castor

Either @poojasingh111, as the data is output as a matrix, you’ll get the same values twice. As the tool compares all variables it’ll compare A to C and also C to A, hence the duplicate. Yes you’re absolutely right - 0 is totally not correlated and 1 or -1 is perfect correlation.

Labels