Convert this k$-$means clustering into the hierarchial based cluserting combination of $2$ and take $2-3$ feature input, since this k$-$means is giving me $66$ percent performance, i want a performance of $90$ percent, write me a step by step code from start till end.