Interpreting Results of Demand Estimation from Machine Learning Models

There is developing interest in the application of Machine Learning Models (MLM) to estimation problems in economics. MLM may be particularly well suited to applications in retail, health care, energy, finance or for web based businesses where large amounts of data are available to help make better decisions and better understand consumer behavior. There are three reasons economists may want to adopt new MLM tools. First is the size of available data sets. Second, these new data sets have many potential predictors where domain knowledge may not be helpful in distinguishing which available data are most relevant. Third, larger data sets allow for modeling more complex relationships than the standard linear model, which is what MLM are able to capture