Close
Type at least 1 character to search
Back to top

COLES

CASE - DATA PROCESSING

Data Factory/

We're proud
of our Work!

CUSTOMER
THE TECHNICAL FRAMEWORK

Oracle Exadata, R, Unix, Control-M Scheduling, DataStage, Microstrategy

SKILLS AND ROLES

1 developer Data engineer

BENEFITS FOR THE CUSTOMER

╋ Development of data modules loading 10 Terabytes of data for fast performance of SQL statements.
╋ 13 external events added to the main table to optimise forecasting model requests: Weather, events and festivals, end date and promotions information products, ....

OUR CHALLENGE

Massive data processing to reconstruct 3 years of sales history with 150 attributes and metrics describing each product in all Coles shops in Australia.

 

THE CONTEXT

Within the Coles data department, we are part of the team building a fast history and large table storing historical sales of each Coles item (product) and external events with 3 years of history. The aim is to forecast demand and optimise the supply chain. Each Coles shop in Australia has a range of items - all products in the shops. The data ingestion team works with the data scientist team to build a fast and massive historical master table. The data scientists train their artificial intelligence models with this table to predict customer demand in each shop.

 

THE PROJECT

╋ Build a master Oracle Exadata table to maximise batch processing and SQL statements and provide detailed data to form Coles AI forecast requests.
╋ Consolidate historical information for 50 million items - merged and curated from DWH data, company operational systems and external sources.
╋ Ingestion of external data sources: obtain from various interfaces the most useful localised information to enrich AI models (e.g. from the Bureau of Meteorology/BOM and for each Australian shop, the weather forecast data from the nearest BOM station).

Our experts

Rahul Kumar Pandey

Julien Labouze