Skip to main content

Table 2 Times in seconds for the fit and predict of each method for all categories

From: Comparison of short-term electrical load forecasting methods for different building types

 

Residential

School

Supermarket

Method

fit

predict

fit

predict

fit

predict

SVR

2.97 ±0.91

0.02

3.21 ±0.73

0.03

11.83 ±1.97

0.04

RF

1.61 ±0.11

0.01

0.6 ±0.05

0.01

3.53 ±0.17

0.01

KNN

0.25 ±0.04

0.12

0.14 ±0.04

0.01

0.22 ±0.05

0.09

MLP

35.74 ±11.03

0.03

157.95 ±43.95

0.03

32.67 ±25.34

0.03

LR

0.12 ±0.01

0.01

0.09 ±0.01

0.01

0.1 ±0.01

0.01

LSTM

529.47 ±274.19

0.13

412.38 ±169.39

0.1

1076.52 ±292.27

0.11

ARIMA

656.81 ±230.55

0.02

1052.88 ±771.1

0.03

18.84 ±11.13

0.01

  1. Mean NRMSE over all test days and all time series. The error is given as standard deviation. The standard deviation of the predict operation is not given, since it is smaller or equal to 0.02 s for all methods and categories