This paper aims to understand to what extent the amount of drug (e.g., cocaine) trafficking per country can be explained and predicted using the global shipping network. We propose three distinct network approaches, based on topological centrality metrics, Susceptible-Infected-Susceptible spreading process and a flow optimization model of drug trafficking on the shipping network, respectively. These approaches derive centrality metrics, infection probability, and inflow of drug traffic per country respectively, to estimate the amount of drug trafficking. We use the amount of drug seizure as an approximation of the amount of drug trafficking per country to evaluate our methods. Specifically, we investigate to what extent different methods could predict the ranking of countries in drug seizure (amount). Furthermore, these three approaches are integrated by a linear regression method in which we combine the nodal properties derived by each method to build a comprehensive model for the cocaine seizure data. Our analysis finds that the unweighted eigenvector centrality metric combined with the inflow derived by the flow optimization method best identifies the countries with a large amount of drug seizure (e.g., rank correlation 0.45 with the drug seizure). Extending this regression model with two extra features, the distance of a country from the source of cocaine production and a country’s income group, increases further the prediction quality (e.g., rank correlation 0.79). This final model provides insights into network derived properties and complementary country features that are explanatory for the amount of cocaine seized. The model can also be used to identify countries that have no drug seizure data but are possibly susceptible to cocaine trafficking.