Long and short time voltage variations are common electromagnetic phenomena that affect the power quality. According to Colombian resolution CREG 024 of 2005, utilities must register for each system busbar and every 10 minutes, PST values per phase, negative sequence unbalance factor V2/V1, number and duration of the long and short-time voltage variations, number and duration of interruptions involved. In this paper, we present the first results of a proposed new method to find the impact of long and short-time voltage variations on the PST and negative sequence unbalance factor indicators, according to colombian regulations and using real feeders measurements, the proposed method uses basic elements of clustering and data mining.