Forecasting customer activities with dilated convolution neural networks: Use case and best practices

This will be presented in English.

Tao Lu (Microsoft), Chenhui Hu (Microsoft)
11:1511:55 Thursday, June 20, 2019
Average rating: ****.
(4.00, 1 rating)

必要预备知识 (Prerequisite Knowledge)

  • A basic understanding of time series forecasting, neural network concepts, and deep learning modeling languages such as Keras or PyTorch

您将学到什么 (What you'll learn)

  • Gain a deep understanding of a customer activities forecasting use case and how to achieve fair comparison with different algorithms
  • Learn best practices for developing a dilated convolution neural network for time series forecasting

描述 (Description)

Forecasting user activities is one of the most common problems that business groups face. Various industries, such as cloud solution providers and online retailers, often need to forecast product consumption and user engagement; the data scientist team under Azure Identity faces the same challenges. With billions of user activities on the Azure cloud, being able to accurately forecast user behavior becomes essential in business planning. However, statistical time series models and machine learning approaches like tree-based models fail to generate forecasts with satisfactory accuracy, while recently developed deep learning models can boost performance, and yet they remain hard to implement.

Tao Lu and Chenhui Hu focus on forecasting customer usage on a daily basis. In this user case study, they explain how Microsoft improved forecasting accuracy by 25% with dilated convolutional neural networks and reduced time in development by 80% with a set of time series forecasting best practices.

Dilated CNNs, recently proposed for modeling sequence data, can achieve state-of-the-art performance in time series forecasting and are easier to train than recurrent neural networks. To speed up the model development process, Microsoft leverages the best practices of time series forecasting that another Microsoft team has created. This is a framework that provides standard workflows of various methods (e.g., statistical methods, traditional machine learning methods, and recently developed deep learning approaches). It also includes reusable utility functions for data standardization and feature engineering. With such utility functions, Microsoft converts the dataset into a standard format. Then, it implements the dilated CNN model and other baseline models by following the templates in the framework. Moreover, the company tunes the hyperparameters of each model with HyperDrive in Azure Machine Learning to achieve a fair comparison of model performance in terms of accuracy, running time, and cost.

Photo of Tao Lu

Tao Lu


Tao Lu is a data scientist in the cloud and AI organization at Microsoft. He has strong background in applying machine learning and deep learning techniques to forecasting problems and deep domain knowledge in cloud identity and the financial services industry. He graduated from University of Washington with a master’s degree in computational finance.

Photo of Chenhui Hu

Chenhui Hu


Chenhui Hu is a data scientist in the Cloud and AI Division at Microsoft. His current interests include retail forecast, inventory optimization, IoT data, and deep learning. He also has research experience in wireless networks and network data analysis. He’s a recipient of the third IEEE ComSoc Asia-Pacific Outstanding Paper Award. He holds a PhD from Harvard University, where his PhD thesis focused on biomedical imaging data mining.

Leave a Comment or Question

Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?

Join the conversation here (requires login)


Picture of Tao Lu
2019-03-05 15:58 CST

Yes we’ll present in English.

011071b6 3e8a9859 | DATA SCIENTIST
2019-03-03 07:50 CST

Would none chinese audience be able to understand the talk?