Customer
Large distribution company, 30+ years on the market, 42 warehouses, 600 suppliers, 4000+ employees.
Challenge
Enabling precise ML-driven forecasting of customer demand in breakdown by key categories of goods and geographic locations for 1M period in order to optimize storage, procurement and logistics.
Project Results
During the project time series forecasting Machine Learning model has been developed, validated and implemented as containerized API-driven service, integrated with Client’s core systems.
Business Value
New models improved existing forecast accuracy for 5-10% depending on the region/group of goods.
Tech Stack
Python (Pandas, Xgboost, Matplotlib), Airflow, Kubernetes, Jenkins, Clickhouse.
Large distribution company, 30+ years on the market, 42 warehouses, 600 suppliers, 4000+ employees.
Challenge
Enabling precise ML-driven forecasting of customer demand in breakdown by key categories of goods and geographic locations for 1M period in order to optimize storage, procurement and logistics.
Project Results
During the project time series forecasting Machine Learning model has been developed, validated and implemented as containerized API-driven service, integrated with Client’s core systems.
Business Value
New models improved existing forecast accuracy for 5-10% depending on the region/group of goods.
Tech Stack
Python (Pandas, Xgboost, Matplotlib), Airflow, Kubernetes, Jenkins, Clickhouse.