China
Africa
Explore chapters and articles related to this topic
Lean in design
Published in Lincoln H. Forbes, Syed M. Ahmed, Lean Project Delivery and Integrated Practices in Modern Construction, 2020
Lincoln H. Forbes, Syed M. Ahmed
Schwaber and Sutherland (2017) developed scrum in the 1990s and define it as a framework within which people can address complex adaptive problems while productively and creatively delivering products of the highest possible value. The word scrum is derived from the game of rugby; a scrum is used to restart activity after a ball has gone out of play. Used since the 1990s, scrum is not a process, technique, or definitive method. Scrum is based on empirical process control theory, Empiricism views knowledge as based on experience and making decisions on what is known. It uses an iterative, incremental approach to assure predictability and minimize risk. Empirical process control is based on three pillars – transparency, inspection, and adaptation. For transparency, there must be a common understanding of significant aspects of the process by all participants, so there can be clarity on the completion or “Done” phase of an activity.
Toward Optimal Variance Reduction in Online Controlled Experiments
Published in Technometrics, 2023
Our approach for count metrics is asymptotically the same as the augmented inverse propensity weighting (AIPW) estimator (Robins, Rotnitzky, and Zhao 1994), whose well-established semiparametric efficiency (Hahn 1998) result forms the basis of our optimality guarantee. For count metrics, we develop valid inference procedures in randomized experiments (L2 convergence in probability to any fixed function), which is much weaker than the pointwise convergence condition to true conditional mean functions as is often required in observational studies (Nichols 2007; Schuler and Rose 2017; Chernozhukov et al. 2018) or the investigation of heterogeneous treatment effects (Chernozhukov et al. 2017; Künzel et al. 2019; Athey and Wager 2019; Nie and Wager 2020; Kennedy 2020), and also differs from the traditional approach of Donsker conditions and empirical process theory (Andrews 1994; Van Der Vaart et al. 1996; Van der Vaart 2000) to control errors in estimating nuisance components (the conditional mean functions in our setting).