Step 1: 
Select 'n' number of measurements to be the centroids (random)

Example: [Gender, Age]
Gender is Cluster 0
Age is Cluster 1

Step 2:
Use the remaining 'n - 126' measurements and group each individual with the random centriod that its more correlated with

Example:

BMI has a higher correaltion with Age => Group with Age (CLUSTER 1)

Step 3: 
For each cluster, find the best centriods

Example:
Gender =>  [Clsuter 0]
	Gender
	Height
	Stride Length

If Gender was centriod for cluster 0, its correlation with Height, and Stride Length = 1.022
If Height was centriod for cluster 0, its correlation with Gender, and Stride Length = 0.944
If Stride Length was centriod for cluster 0, its correlation with Height, and Gender = 1.0005

Gender has a higher total correaltion => its the best cnetriod 
Repeat for every other cluster

Step 4:
With the new centriods, repeat step 1 - 4 until the centriods don't change

Example: 
Best Centriod for cluster 0 was Gender and best centriod for cluster 1 is BMI 
 	Repeat steps until these centriods stay the same