Dear Madam

Maintainability can be measured by Maintainability Index ( MI) which a software metric which measures how maintainable (easy to support and change) the source code is.

MI is dependent on

 Lines Of Code, Cyclomatic Complexity and Halstead volume.

So we must have

V = Halstead Volume

G = Cyclomatic Complexity

LOC = count of source Lines Of Code (SLOC)

CM = percent of lines of Comment (optional)

From these measurements the MI can be calculated:

MI = 171 - 5.2 * ln(V) - 0.23 * (G) - 16.2 * ln(LOC)

So in K-means algorithm, all above may be the features or factors of vector of MI.

 #Dinesh

Dear Bhawana Mathur, you may use K-means algorithm for maintainability prediction, but before applying this clustering, you should define the range of MI. Because K-Means always give district value not continuous value.

Dear Dinesh,  we can't use V, G, LOC, CM as feature for clustering, because you are using these feature to calculate MI index based on halsteded equation. You may consider some other metrics such as WMC, DIT, NOC, CBO etc as features for prediction.

WMC, DIT, NOC, CBO are the best features