Introduction

The Client is an American department store retail chain, which is also the largest omnichannel retailer in America. They have operated over 50 years and continue to provide wonderful  products and remarkable savings to their customers. The client is mainly focused on building unique shopping experiences and increasing customer satisfaction by providing great deals and customized personalized suggestions.

Why We Chose Pluto7

Being in the business for more than 5 decades translated into a wide range of products being offered by the client made it difficult to get a highly accurate forecast of demand patterns for each and every product. The biggest challenge here was that the demand for new products or the products with limited historical data could not be predicted by existing off the shelf solutions. Therefore, the client looked for a Google Cloud Partner that can customize the solution as per their need and deliver accurate predictions in spite of inconsistencies in the data.

The Client partnered with Pluto7 because of their proven expertise in the retail and supply chain domain, Google Cloud Platform, and its various components including AI and Machine Learning. With their experience of obtaining high forecast accuracies for other enterprises using machine learning, Pluto7 has been endorsed by Google for offering top-notch demand forecasting solutions. Further, our team leveraged Google Cloud Platform and Pluto7’s Demand ML solution to help the client’s team predict and forecast demand accurately. 

Solution

The Solution consists of a ML model that would predict the sales of all the client offerings using a combination of historical sales data and many other internal and external sources. Pluto7 team leveraged Demand ML to perform exploratory data analysis.

Additionally, Google AI Platform was used for training existing and exploring new ML models to get the right accuracy. The model chosen generated a highly-accurate forecast for all the subclasses within the run window of 48 hours, using a mechanism called High Throughput Computing (HTCondor). Further the generated results were presented through visuals over datastudio. 

Results

The demand ML solution was successfully scaled up to meet the specific client’s needs, delivering 85% accuracy across all products. Pluto7 built and deployed a top-performing ML model yielding highly accurate results using historical, other predictive trend data, as well as weather information and more. The client then expanded ML usage from predicted demand for supply and price planning as well. 

Industry Retail

Solutions Demand-ml

Challenges

  • Client needs reliable forecasts across all the product segments and skus with explainability.
  • Handling new line products and the products with limited history was tricky
  • Scaling up the Demand ML Solution(built by Pluto7) for a reliable forecast demand with the right architecture is key.

Results

  • The Hybrid ML Model gained more than 85% of accuracy across all the subclasses.
  • The SLA to scale over all the products within 48 hours was achieved through High Throughput Computing and the actual time taken was around 8-10 hours.
  • The forecasted demand was further used to calculate the supply and to perform price planning as well.

Products Used

  • Google Cloud Platform
  • Google Bigquery
  • Google Cloud Storage Bucket
  • Google Cloud AI Platform
  • HTCondor
  • Data Studio

Customer Success Stories

  • Success Story: Pluto7 helped a major service provider to cut down infrastructure cost by 20%

    Read more
  • Success Story: Pluto7 helped a leading Healthcare Analytics Company in extracting entities with 90% accuracy

    Read more
  • Success Story: Pluto7 helped a leading Singapore based cloud-based phone solutions provider improve the quality of audio files

    Read more

Talk to an Expert

Transform your business by leveraging the power of Machine Learning Artificial Intelligence, Analytics, and IoT solutions.

Contact Us