Skip to main content

Table 3 Predictive comparisons of models based on the sum of the log-likelihood

From: Detecting local risk factors for residual malaria in northern Ghana using Bayesian model averaging

Training Testing Sum of log-likelihood
Logistic Lasso BMA
Base model (p = 29)a
 Rainy 2010 Rainy 2011 − 1072.27 − 1049.24 − 1028.24b
 Rainy 2011 Rainy 2012 − 1055.39 − 1037.49 − 1032.57b
 Rainy 2010 Rainy 2012 − 1153.62 − 1110.05 − 1057.07b
 Dry 2011 Dry 2012 − 969.88 − 919.45b − 921.54
 Dry 2012 Dry 2013 − 915.95 − 897.03b − 903.60
 Dry 2011 Dry 2013 − 967.83 − 920.28 − 915.81b
  Average − 1022.49 − 988.92 − 976.47
Model with interactions and splines (p = 73)a
 Rainy 2010 Rainy 2011 − 1079.63 − 1042.02 − 1027.85b
 Rainy 2011 Rainy 2012 − 1066.56 − 1035.44 − 1030.75b
 Rainy 2010 Rainy 2012 − 1156.76 − 1092.52 − 1050.55b
 Dry 2011 Dry 2012 − 1065.27 − 1029.66 − 921.05b
 Dry 2012 Dry 2013 − 922.40 − 902.79 − 902.32b
 Dry 2011 Dry 2013 − 1079.34 − 1059.24 − 917.82b
  Average − 1061.66 − 1026.95 − 975.06
  1. BMA Bayesian model average
  2. ap refers to the number of covariates in the model
  3. bIndicates the model with the best fit