Enterprise Data Analytics For Asset Repair/Replace Decision-Making

Client

One of the largest retailers that operates a chain of hypermarkets, discount department stores, and grocery stores around the globe

Industry

Retail

Services

Product Development, Web, DevOps, Cloud

Tech

Azure DevOps pipelines, Terraform, Azure CLI, PowerShell, Docker Python, Spark, Delta Lake, Airflow, Docker, Kubernetes, Java 11, Spring Boot, REST API, OpenAPI, Maven, Spring Data (JPA), OAuth 2.0, MS SQL, Azure (App Services, Load Balancers, Key Vault, Application Insights)

Challenge

In large enterprises, asset management decisions to repair or replace a “bad” asset approaching the final phase of its life cycle present a big challenge. The wrong choice can directly affect profitability, leading to downtime or unnecessary costs. If you replace an asset too soon, you are wasting resources, but if you delay too long choosing repairs instead, operations can be put at risk. Sound asset management was a natural pain point, too, for our client, a leading retailer operating thousands of stores in many countries. They needed to know precisely what was the best course of action for their vast assets to improve cost efficiency. The retailer approached ITRex, its long-standing tech partner in big data services, for assistance with building an enterprise-grade data analytics solution for ML-powered decision-making.

Our task was to:

●

Build a training data set and prepare the data to train the client’s proprietary ML model

●

Collect, structure, clean and enrich data on assets and stores that operate them

●

Build the entire back end of the enterprise data analytics solution, pairing the ML model to the big data

●

Ensure that the ML model can be retrained to improve predictive performance

●

Build an easy-to-understand data analytics dashboard for end-users

Solution

An enterprise data analytics solution running on a cloud-based architecture has been built using DevOps best practices. By analyzing multiple inputs about a specific asset in a specific store and running cost calculations, the system allows both data analysts and non-tech savvy business users to get an instant answer about the cost efficiency of repairing vs. replacing this asset. Specifically, the system:

●

Allows users to select an asset from a table of attributes for analysis

●

Uses a predefined set of dozens inputs for cost calculations, from age, useful life, initial capital cost, book value, and critical repair cost to current resale value

●

Allows users to customize the inputs

●

Retrieves and visualizes data on historical and forecast maintenance and repair costs

●

Runs ML-powered calculations and displays a comparison of net present value for repair and replace scenarios, recommending which scenario is best

●

Displays detailed maintenance and repair cost breakdown

●

Retrains the ML model based on fresh asset data