Implementation of Process-Based and Data-Driven Models for Early Prediction of Construction Time

Table 1

Model accuracy results for BTC-GRNN model for (a) training data and (b) validating data (using 10-fold cross-validation) [63].

(a) Training data
Mean target value for input data	5.1452765
Mean target value for predicted values	5.1456292
Variance in input data	1.1319931
Residual (unexplained) variance after model fit	0.0031972
Proportion of variance explained by model (R^2)	0.99718 (99.718%)
Coefficient of variation (CV)	0.010989
Normalized mean square error (NMSE)	0.002824
Correlation between actual and predicted target values	0.998588
Maximum error	0.2234627
RMSE (root mean squared error)	0.0565438
MSE (mean squared error)	0.0031972
MAE (mean absolute error)	0.0310945
MAPE (mean absolute percentage error)	0.6133842
(b) Validation data
Mean target value for input data	5.1452765
Mean target value for predicted values	5.1156255
Variance in input data	1.1319931
Residual (unexplained) variance after model fit	0.0772845
Proportion of variance explained by model (R^2)	0.93173 (93.173%)
Coefficient of variation (CV)	0.054030
Normalized mean square error (NMSE)	0.068273
Correlation between actual and predicted target values	0.965664
Maximum error	1.2041075
RMSE (root mean squared error)	0.2780008
MSE (mean squared error)	0.0772845
MAE (mean absolute error)	0.1727287
MAPE (mean absolute percentage error)	3.34113