Welcome to the course
Welcome to our course, Business Analytics Using Forecasting. I’m Galit Shmueli, a professor at National Tsing Hua University in Taiwan. I’ve been designing and teaching business analytics courses for over a decade at University of Maryland (USA), Indian School of Business (India), NTHU (Taiwan), and Statistics.com. During this six-week course I aim to help you understand not just how to crunch time series data but why to crunch data for creating forecasting solutions.
People, societies, products and processes generate lots of data. Data science and business analytics have evolved as more and more data becomes available at higher resolution and at higher frequency. Quantitative forecasting is the science of using time series data for generating forecasts. In other words, extrapolating a series of measurements into the future. Forecasting plays an essential role in decision making in almost any environment you can imagine where data are collected over time.
Course topics and pace
The course will introduce you to the different steps in the forecasting process, from defining a forecasting goal, to developing and evaluating forecasting models and solutions.
- Week 1 focuses on different types of goals that call for forecasting, and link the business goal to forecasting concepts such as forecast horizon, description/prediction, automation, delays, and forecast updating.
- Week 2 is about exploration and visualization for time series data. We will look at pattern types common in time series and how to identify them. This is where we showcase the power of interactive visualization tools.
- In Week 3 we delve into performance evaluation, a critically important component of forecasting. We’ll learn how to set up the data for proper evaluation, what to measure, and how to compute, display, benchmark, and evaluate forecasting performance.
- Week 4 introduces a popular family of forecasting methods called smoothing. They include methods such as the moving average, simple exponential smoothing, and advanced exponential smoothing. We’ll see when it is appropriate to use different smoothing methods, and why.
- In Week 5 we discuss the use of linear regression models and how to capture different trend and seasonality patterns.
- Week 6 expands the discussion of regression for capturing the important pattern of autocorrelation (the relationship between values in neighboring periods). We also look at how to integrate external data into our forecasting model. Finally, we talk about big data and the Internet-of-things in the forecasting context.
While the course is not mathematical in nature, you’ll find that weeks 4-6 gradually become a little more technical. We provide multiple examples and discussion opportunities to help you clarify doubts and evaluate your knowledge.
Learning from each other
Please take the time to join in the activities and discussions during the course. The comments areas will be a valuable place to share your experience and post questions. Do try to help other learners if you can. The course mentors will try to advise on common queries - please meet Mahsa Ashouri, Viet-Cuong (Daniel) Trieu and Travis Greene - you might like to ‘follow’ them so you can more easily find advice they have given in response to queries. Please note that they will not be able to read and reply to all of your comments though.
Course materials and resources
All the materials required for the course will be available on the course website: videos, articles, quizzes, discussions, tests. For datasets and external resources we will provide links. All the datasets are at http://www.forecastingbook.com/mooc.
If you’re interested in getting the textbook associated with this course (it is not a requirement, but can be very helpful), see the website for Practical Time Series Forecasting: A Hands-On Guide for purchasing options for the R and XLMiner editions.