Table 5

Results for comparison of elastic net (ENET) for zero-inflated Poisson and ‘traditional’ predictors selection methods

Full modelBackward
stepwise
regression
ENET without cross-validationENET with cross-validation
Akaike information criterion (AIC)666.6665.23662.31
Bayesian information criterion (BIC)909793750
Log likelihood−929−301.7−298.0−46.65
  • For AIC and BIC, lower values are indicative of better model performance while for log likelihood, a larger number is indicative of better model performance.

  • Calculation of AIC and BIC is not possible with ENET with cross-validation.