The smart grid is a critical infrastructure area, so machine learning models involving it must be interpretable in order to increase user trust and improve system reliability. This research work concludes that the highest accuracy value is 88.02% of accuracy level which is acquired by n-neighbors =5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The highest positive predictive value is 0.87 which is owned by n-neighbors=5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The maximum as well same Sensitivity value is 0.88 which yielded by n-neighbors=3 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model, n-neighbors=5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. n-neighbors=5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The highest F-Measure value is 0.86 is produced by n-neighbors=3 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model, n-neighbors=5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model, n-neighbors=7 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The supreme value of Mean Square Contingency Coefficient shows that 0.45 which is given by n-neighbors=5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The maximum amount of kappa value is 0.42 which is picked by n-neighbors=3 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The highest ROC value is 0.80 which is given by n-neighbors =9 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. Highest PRC value is 0.87 which is given by there are two models. They are n-neighbors=7 and K-9 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function models. The Least Mean absolute deviation is 0.16 which is given by is shown by n-neighbors=1 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The least root mean squared deviation is 0.31 which are demonstrated by n-neighbors=5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The smallest relative absolute error performance is 61.66 %which is hold by n-neighbors=1 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. The largest root relative squared error performance is 110.98 % which is hold by n-neighbors=1 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model. This research work recommended that the n-neighbors =5 with 1.0 of default radius,30 of leaf size, minkowski of distance metric, Ball Tree of Nearest Neighbor Searching algorithm, without parallel job, and Manhattan distance function model due to its performance is best compare with other models with low deviations.