BexExpress: Automation of address recognition from manual input
Case Study
Our client needed to create a script that will automate the process of recognising addresses that their system can’t. Script was required to run fast because it will work on tens of thousands of addresses.
Our Client
BexExpress is the biggest private courier company in Serbia.
The Tech Stack
Database
The initial data was imported into SQLite, a lightweight and fast relational database. It provided an efficient storage solution for handling the data during the initial stages of the project, offering easy integration and setup.
Data Processing
Once the data was in SQLite, we utilized Python for processing and transforming the data. Python’s powerful libraries and flexibility made it ideal for cleaning, manipulating, and analyzing the dataset, ensuring the information was ready for further use or integration into other systems.
Problem
Manual input is unpredictable, options that someone can write are limitless. We needed to find a solution how we will predict what the user wrote with 100% certainty that we found a match in a client’s database of addresses.
Solution
Creating multiple algorithms that parse the given string, work on it and try matching it with street and municipality from a database. Optimised to work fast and precisely. If the program doesn’t guarantee, the match string is forwarded for manual validation.
Conclusion
Using our solution client saves hundreds of hours every month. We managed to automate validation of 95% of cities and two thirds of streets and allow BexExpress to not only save time, but also ship faster and scale their operations.
