TheK-Nearest Neighbor classifier is a well-known and widely applied method in datamining applications. Nevertheless, its high computation and memory usage costmakes the classical K-NN not feasible for today’s Big Data analysisapplications. To overcome the cost drawbacks of the known data mining methods,several distributed environment alternatives have emerged. Among thesealternatives, Hadoop MapReduce distributed ecosystem attracted significantattention. Recently, several K-NN based classification algorithms have beenproposed which are distributed methods tested in Hadoop environment andsuitable for emerging data analysis needs. In this work, a new distributedZ-KNN algorithm is proposed, which improves the classification accuracyperformance of the well-known K-Nearest Neighbor (K-NN) algorithm by benefitingfrom the representativeness relationship of the instances belonging todifferent data classes. The proposed algorithm relies on the data classrepresentations derived from the Z data instances from each class, which arethe closest to the test instance. The Z-KNN algorithm was tested in a physicalHadoop Cluster using several real-datasets belonging to different applicationareas. The performance results acquired after extensive experiments arepresented in this paper and they prove that the proposed Z-KNN algorithm is acompetitive alternative to other studies recently proposed in the literature