Bayesian linear model with zero-inflated covariates

Covariates sometimes called predictors or explanatory variables are factors that explain or predict the response or dependent variable. The impact of such covariates can be tested using several statistical tools depending on the nature of the data under consideration. This study was carried out to measure the distribution of zero-inflated covariates by fitting Zero-inflated Poisson, Zero-inflated Generalized Poisson and Zero-inflated Negative Binomial at different levels of covariates and sample size. Zero-inflated Poisson outperformed other models when the sample size was high while Zero-Inflated Negative Binomial tends to perform better at smaller sample size. The choice of which model to consider depends on the size of sample, proportion of zeros and the number of covariate to be included in the model. However, ZIP and ZINB look more suitable under any of the aforementioned scenario.