Skip to main content

Table 2 Results of applying multinomial logistic regression to three sample applications and three groups of environmental predictor variables

From: Prediction of benthic community structure from environmental variables in a soft-sediment tidal basin (North Sea)

Group of predictors H hydrodynamics S sediment H + S hydro + sedim.
Application A1: spatial interpolation
 Best predicting set of covariates, based only on training set Dry, τ meanc , (τ maxw )2 logD10 Dry, τ meanc , (τ maxw )2, logD10
 CCR for training set 0.63 ± 0.030 0.50 ± 0.049 0.68 ± 0.031
 CCR for test set 0.61 ± 0.031 0.49 ± 0.058 0.64 ± 0.031
 Cohen’s kappa for training set 0.47 ± 0.043 0.26 ± 0.079 0.54 ± 0.046
 Cohen’s kappa for test set, κTEST 0.44 ± 0.044 0.25 ± 0.077 0.49 ± 0.044
Application A2: extrapolation to a spatial gap in the benthos data
 Best predicting set of covariates, based only on training set Dry, τ meanc , (τ maxw )2 logD10, logD50 Dry, τ meanc , (τ maxw )2, logD50, (logD50)2
 CCR for training set 0.65 ± 0.018 0.53 ± 0.028 0.71 ± 0.017
 CCR for test set 0.58 ± 0.034 0.42 ± 0.051 0.62 ± 0.054
 Cohen’s kappa for training set 0.50 ± 0.026 0.29 ± 0.043 0.58 ± 0.025
 Cohen’s kappa for test set, κTEST 0.40 ± 0.051 0.14 ± 0.077 0.44 ± 0.077
Application A3: extrapolation to a new drainage area
 Best predicting set of covariates, based only on training set Dry, τ meanc logD10, logD50, logD90 Dry, dry2,
 τ meanc , logD50
 CCR for training set 0.67 ± 0.032 0.62 ± 0.028 0.73 ± 0.021
 CCR for test set 0.53 ± 0.041 0.51 ± 0.043 0.51 ± 0.038
 Cohen’s kappa for training set 0.52 ± 0.047 0.42 ± 0.049 0.61 ± 0.030
 Cohen’s kappa for test set, κTEST 0.34 ± 0.048 0.25 ± 0.053 0.26 ± 0.045
  1. The results of application A1 are averages and standard deviations of 200 different (random) subdivisions of the full dataset. The standard deviations of applications A2 and A3 were calculated from 200 bootstrap replications of the training set. CCR is the “correct classification rate”