PinnedShannon KelleyHow to Develop Data Science Projects for Production DeploymentsAn abridgment of Pete Cacioppi’s “Tidy, Tested, Safe” with DevOps extensions for deploying data science applicationsAug 29Aug 29
PinnedShannon KelleyHow to Implement Column Generation for Vehicle RoutingIntuition behind and an implemented example of how column generation can be used to speed up solution methods for vehicle routing problemsSep 7, 20225Sep 7, 20225
Shannon KelleyHow to Guarantee a Data Science Model Runs for Business UsersSolution: Serve it as a web app from a container.Aug 29Aug 29
Shannon KelleyHow to Guarantee a Data Science Model Runs for DevelopersSolution: Host it on a package manager.Aug 29Aug 29
Shannon KelleyHow to Guarantee a Data Science Model Runs for CollaboratorsSolution: Define a virtual environment.Aug 29Aug 29
Shannon KelleyHow to Guarantee a Data Science Model Runs in any EnvironmentSolution: Package the model. Then automate set up by defining a virtual environment, hosting on a package manager, and/or serving as a web…Aug 29Aug 29
Shannon KelleyHow to Integrate Contributions to a Data Science Model from Many DevelopersSolution: Use a version control system (VCS) and a branch flow strategy.Aug 29Aug 29
Shannon KelleyHow to Keep Data Science Models from CrashingSolution: Assert each function’s assumptions on inputs are met and instruct how to fix violations.Aug 29Aug 29
Shannon KelleyHow to Keep Data Science Model Results Consistent across ReleasesSolution: Write unit tests.Aug 29Aug 29
Shannon KelleyHow to Keep Data Science Models from Giving Funny ResultsSolution: Use an Object Relational Mapper (ORM) to define model inputs, enforce their adherence, and instruct how to resolve violations.Aug 29Aug 29
