Abstract: In this session, we will talk about innovation in traditional financial planning. We developed a machine learning process to produce quarterly and annual goals for the company. The business question is ""if we do X, what would happen in Y"". In Economics and Machine learning, this type of question is called ""Causal inference"" to predict an outcome after an action. In Finance, the company has goals for certain top business metrics such as revenue and number of users, and each business team (marketing, sales, etc.) will be responsible to achieve goals.
The causal inference process will automatically look for factors that can lead to moving a business metric. In our example, one of the top primary goals is to increase revenue. Senior executives want to increase revenue by moving the number of monthly active users (MAU), the cost per acquisition, and several other leadership metrics. If we have a goal of increasing revenue by 20%, how many more MAU should we have? This type of causal question is answered by causal models.
Causal models are not just linear regression. They are machine learning-based models to control other variables as fixed numbers to observe the impact between X and Y. During the session, we not only provide the business problem and results, but also want to provide practical advice in our modeling process. Lastly, we will share unique experience of working with cross-functional leadership from product, marketing, finance, and operations.
Bio: Jeremy Wenxiao Gu is the Director of Data Science and manages the statistical experimentation team at Shipt Inc, San Francisco. The company is an American delivery service owned by Target Corporation. Before Shipt, Jeremy was Data Science Manager at Stitch Fix (2020-2021), Data Scientist and Manager at Uber (2017-2019), and Data Scientist at Amazon (2014-2017). Jeremy received an MS in Statistics from the University of Washington (2014) and a BS in Mathematics and Statistics from the University of Minnesota (2012). Jeremy is also a member of the American Statistical Association (ASA). He served as Chapter Vice President and Chapter Representative of ASA for three years (2015-2018).