class: center, middle, inverse, title-slide # swissdata v1.0 ## Learnings from the Past & Roadmap for the Future --- class: center, middle ## Motivation ### Why Revamp swissdata ? <div class="my-footer"><span><div class="footer-text" align="right">These slides and video presentation by Dr. Matthias Bannert are licensed under <a href="https://creativecommons.org/licenses/by-nc-sa/4.0/?ref=chooser-v1">CC BY-NC-SA 4.0</a></div></span></div> --- ## 1. New FSO .csv + ODS files <img src="fsocsv.png" width="800" /> - easier to read maintain - metadata kept separate by FSO already - heterogenerous across datasets, but a step towards swissdata format --- ## 2. First Level Support Has Been too Hard to Split <img src="contributors.png" width="400" /> --- ## 3. Timing Plenty of Funding Opportunities and Partners Available: - https://github.com/statistikZH/statbot (collab with FSO, DataShop Kanton ZH, ZHAW) - Open Science and Open Research Data calls from ETH, Swiss Data Science Center, Swiss Universities - FSO very cooperative, in particular w.r.t. to macroeconomic data --- class: center, middle ## Goals --- ## 1. Easy Debugging Interactive Use --- ## 2. Facilitate Automation / Production Use --- ## 3. Improve Expandability - Build Community --- class: center, middle ## Implementation Strategy --- ## 1. Modularize - swissdata.core (engine, skeleton/boilerplate, read, write, keep history in folder etc.) - swissdata.snb, swissdata.fso (pacmans) (separation of concern, integration of data already in sd format (SECO) and data that needs to be converted) --- ## 2. 5-Star Documentation - Project Documentation (swissdata, paired file approach) - pkgdown for swissdata.core, swissdata.snb etc. - common documentation, inheritParams etc. --- ## 3. Improved Boilerplates - function skeletons - more transparent dispatch - validate functions are key - guidance in case of dataset issues, help track 'em down --- class: center, middle ## Roadmap --- ## Milestones --- ## Gantt-Chart