Hi@Pushpender,
You can do that. But K-Means is not used for this. You need to create the clusters and discard all the dataset those are not giving the perfect graph. But it is time-consuming and if it fits with your use case, then only you can plan in this way.