Accurate identification of people or places of interest is essential for driving business decisions and enhancing government services. A critical part of this process is accurate geocoding, which relies heavily on the completeness of address data. In many cases, street addresses lack country information, leading to poor geocoding accuracy and suboptimal results. Specifying the country significantly improves both accuracy and performance during geocoding.

This deep learning model addresses this challenge by automatically predicting the country for incomplete addresses. Trained on address data from openaddresses.io, the model is capable of classifying addresses from 18 different countries, enabling more precise geocoding and location-based analysis.