Finding an accurate CBL estimation method for residential customers became an interesting topic in recent years. One of the promising approaches suggested in literature is clustering the customers in order to eliminate their randomness and gaming incentives. If you know any clustering methods that can be helpful in this endeavor, please recommend it in this thread, and explain why do you think the proposed clustering method can address the volatility of residential customers?
I think you may use "Hopkins statistics clustering" in order to handle randomness of data.
You may follow the below links....
http://ieeexplore.ieee.org/document/1375706/
http://datascience.stackexchange.com/questions/14142/cluster-tendency-using-hopkins-statistic-implementation-in-python
@Sanjay your suggestion is really helpful. I appreciate your suggestion.
I think that you must also look the metaheuristic approaches methods like Genetic Algorithm, Ant Colony Optimisation, Classical Mechanics Optimization. Because this approches are an artificial intelligence methods to solve the combinatories problems.
@Charaf Meta-heuristic approaches like what you mentioned are for optimization. In this case, we want to create a suitable control group. In order to do so, we need to have a good clustering method to be able to group similar customers together and create a good cluster.