Preprocessing Approach for Power Transformer Maintenance Data Mining Based on k-Nearest Neighbor Completion and Principal Component Analysis

Table 4

Statistical data after imputations.

Imputation by KNN (k = 5)						Imputation by mean		Imputation multiple
Variable	Obs	Obs. with MV	Obs. without MV	Mean	Standard deviation	Mean	Standard deviation	Mean	Standard deviation
H2	31	0	31	80,323	18,734	80,034	18,665	79,064	19,062
CH4	31	0	31	30,374	10,074	30,387	10,074	30,291	10,088
C2H4	31	0	31	30,819	10,832	31,080	10,734	30,772	10,870
CO	31	0	31	1174,903	293,062	1174,400	293,049	1184,598	298,499
CO2	31	0	31	9045,484	3072,597	9159,467	3006,342	9173,076	3007,297
O2	31	0	31	5591,968	5126,700	5591,968	5126,70	5591,968	5126,700
N2	31	0	31	71286,032	10399,787	71206,679	10264,423	71291,446	10547,466

MV: missing value.