Sep 23–26, 2019

Time Series Forecasting using Deep Learning with PyTorch

Ying Yau (AllianceBernstein)
11:20am12:00pm Wednesday, September 25, 2019
Location: 1A 06/07
Secondary topics:  Deep Learning, Financial Services, Temporal data and time-series analytics

Who is this presentation for?

Data Scientist, Statisticians, Data Science VP/Director/Manager, Anyone who needs to conduct time series forecasting




Forecasting is both a fascinating subject to study and an important techniques applied in industry, government, and academic settings. Example applications include demand and inventory planning, marketing strategy planning, capital budgeting, pricing, machine predictive maintenance, macroeconomic forecasting, and supply chain forecasting.

Forecasting typically requires time series data, and time series data is ubiquitous nowadays, both within and outside of the data science field: weekly initial unemployment claims, tick-level stock prices, weekly company sales, daily number of steps taken recorded by a wearable, machine performance measurements recorded by sensors, key performance indicators of business functions, just to name a few.

However, time series data differs from cross-sectional data in that time series data has temporal dependence and this dependence can be leveraged to forecast future values of the series. Some of the most important and commonly used data science techniques to analyze time series data and make forecast based on them are those in developed in the field of statistics and machine learning. For this reason, time series statistical and machine learning models should be included in any data scientists’ toolkit.

This session discusses the application of neural network-based techniques to time series forecasting problems and compares them to time series statistical models when modeling and forecasting time series with trends, multiple seasonality, regime switch, and exogenous series. Specifically, it will discuss how neural network-based models can be used to modeling and forecasting jointly multiple forecasting time series that may have trends, multiple seasonality, and potential regime switch, and how they can be used to incorporate exogenous time series. I will compare the characteristics of neural network-based techniques to the popular statistical models such as the class of Vector Autoregressive (VAR) Models and Regime Switching Models.

I will discuss the advantages and disadvantages when using each of these models in time series forecasting scenarios. Real-world applications, such as dynamic asset allocation, are used to illustrate these techniques. While not the focus in this presentation, exploratory time series data analysis will also be included in the presentation.

This presentation is suitable for data scientists who have working knowledge of the classical linear regression model and a basic understanding of univariate time series models, such as the class of Seasonal Autoregressive Integrated Moving Average Models and machine learning techniques.

Prerequisite knowledge

Working knowledge of Python Good understanding of classical linear regression models, basic machine learning techniques (SVM, Tree-based models, and neural network based models) Some knowledge of time series forecasting

What you'll learn

The attendees will learn The mathematical formulation and coding implementation of Neural Network-Based Models using PyTorch The mathematical formulation, key characteristics, and coding implementation of Vector Autoregressive (VAR) Models framed using the State Space Formulation and Regime Switching Models The pros and cons of the application of Neural Network-Based models, VAR, and Regime Switching models when modeling and forecasting time series with trends, multiple seasonality, and regime switch
Photo of Ying Yau

Ying Yau


Jeffrey is a Distinguished Data Scientist at WalmartLabs, where leads data science for the store technology department. His prior roles include the Chief Data Scientist at AllianceBernstein, a global asset-management firm, Vice President and Head of Data Science at Silicon Valley Data Science, and senior leadership position at Charles Schwab Corporation and KPMG. He has also taught econometrics, statistics, and machine learning at UC Berkeley, Cornell, NYU, University of Pennsylvania, and Virginia Tech. Jeffrey is active in the data science community and often speaks at data science conferences and local events. He has many years of experience in applying a wide range of econometric and machine learning techniques to create analytic solutions for financial institutions, businesses, and policy institutions. Jeffrey holds a Ph.D. and an M.A. in Economics from the University of Pennsylvania and a B.S. in Mathematics and Economics from UCLA.

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)

Contact us

For conference registration information and customer service

For more information on community discounts and trade opportunities with O’Reilly conferences

For information on exhibiting or sponsoring a conference

Contact list

View a complete list of Strata Data Conference contacts