Supervise learning algorithms, such as regression analysis, support-vector machines, and neural networks have demonstrated phenomenal performance in the era of big data. However, they often fail in answering the question, what would happen if the world changed in some specific way while holding other variables fixed? Such problems arise in many business applications including in finance, policymaking, and healthcare. This course covers modern econometric techniques for evaluating causal effects based on observational (that is, non-experimental) data. Topics covered in the course include multivariate linear regression, heteroscedasticity and weighted least squares (WLS), dummy variables and interactions, difference in differences (DD), logistic regression, probit model, censored regression models, exact matching, propensity score matching (PSM), regression discontinuity design (RDD), fuzzy regression discontinuity (FRD), synthetic control, instrumental variables (IV), and two-stage least squares (2SLS). Students get hands-on experience using R.