Background
Product recommendation systems for retail e-commerce businesses suggest popular products for new customers. Recommendations are then updated to other products based on purchase history and ratings provided by other customers. As the customer purchase history grows over time, the retailer can bring more targeted offers to the customer.
Analyzing the rich datasets for in-store and online spending behavior helps to reveal profit potential while considering various retailer details, such as geography, sector, or size.
This AI reference kit uses Intel® AI software, optimized for performance, to help retailers better predict future consumer purchases based on prior purchase behavior faster. Retailers can improve purchase prediction efficiency and accuracy, which can result in increased customer acquisition, retention, and to maximize cross and upselling opportunities.
Solution
In collaboration with Accenture*, Intel developed this purchase prediction reference kit, which may assist your application to increase efficiency, accuracy, and speed of predicting purchase behavior to help scale customer retention and sales volume. This reference kit includes:
- Training data
- An open source, trained model
- Libraries
- User guides
- Intel AI software products
At a Glance
- Industry: Retail
- Task:
- Preprocess data to determine product category and customer segment categories
- Train and inference three different classifier models in each category
- Dataset: E-commerce purchase history
- Type of Learning: Semisupervised learning followed by supervised learning
- Models: K-nearest neighbors and random forest classifiers
- Output: Predict customer purchases based on the customer segment
- Intel AI Software Products:
- Intel® AI Analytics Toolkit (AI Kit)
- Intel® Extension for Scikit-learn*
Since purchase prediction can be a compute-intensive operation for inference workloads because of the large dataset sizes, this experiment illustrates how Intel software products speed up model training and inferencing with greater accuracy.
Technology
Optimized Intel® AI Software Products for Better Performance.
- AI Kit—Achieve end-to-end performance for AI workloads.
- Intel Extension for Scikit-learn—Help enable faster and more effective training models. A better trained model can improve the accuracy of purchase predictions.
Performance was tested on Microsoft Azure* Standard_D4_v5 using 3rd generation Intel® Xeon® processors for optimized performance.
Benefits
To build a purchase prediction application for a retail organization, developers build and train models using customer purchase history. Optimized for performance with Intel software products that enable faster training and inference, this kit helps the retailer deliver targeted offers to customers to help with cross and upselling opportunities.
