Rank correlation is when two variables are ranked the change in one shows the same/positive/negative change in another rank when we measure it across two points. Don’t worry if you still don’t understand, we will find Kendall rank correlation using below dataset.
We are trying to see if there is any correlation if size of engine increases will it affect the price of the car.
You can download dataset here
For Kendall rank correlation we need to arrange our dataset to ascending order as shown below.
Now let’s assume Engine size as X and price as Y, now we need to find a pair to check the Rank correlation.
As per our data set above we take X1 and X6 observation to see the Rank correlation
X1=108 and Y1=16430
X6=131 and Y6=23875
When we compare X1 to X6 and Y1 to Y6, we know X1 < X6 and Y1 < Y6 we call this scenario as concordant pair.
Now if we compare X1 with X4
X1=108 and Y1=16430
X4=130 and Y4=13495
When we compare X1 to X4 and Y1 to Y4, we know X1<X4 and Y1 > Y4 we call this scenario as discordant pair.
If you substitute the values the result will be 0.459619.
1)A correlation coefficient of 1 means that for every positive increase in one variable/dataset, there is a positive increase of a fixed proportion in the other. For example, shoe sizes go up in (almost) perfect correlation with foot length.
2)A correlation coefficient of -1 means that for every positive increase in one variable, there is a negative decrease of a fixed proportion in the other. For example, the amount of gas in a tank decreases in (almost) perfect correlation with speed.
3)Zero means that for every increase, there isn’t a positive or negative increase. The two just aren’t related.
Data and Machine by viswateja 