Workshop on FAIR and Frictionless workflows for tabular data

4TU.ResearchData and Fric­tion­less Data joined forces to orga­nize the work­shop “FAIR and fric­tion­less work­flows for tab­u­lar data”. The work­shop took place on 28 and 29 April 2022 in an online for­mat

On 28 and 29 April we ran the work­shop “FAIR and fric­tion­less work­flows for tab­u­lar data” in col­lab­o­ra­tion with mem­bers of the Fric­tion­less Data project team. 

This work­shop was envi­sioned as a pilot to cre­ate train­ing on repro­ducible and FAIR tools that researchers can use when work­ing with tab­u­lar data, from cre­ation to pub­li­ca­tion. The pro­gramme was a mix­ture of pre­sen­ta­tions, exer­cis­es and hands-on live cod­ing ses­sions. We got a lot of inspi­ra­tion from The Car­pen­tries style of work­shops and tried to cre­ate a safe, inclu­sive and inter­ac­tive learn­ing expe­ri­ence for the par­tic­i­pants.

The work­shop start­ed with an intro­duc­tion to Repro­ducible and FAIR research giv­en by Eiri­ni Zorm­pa (Train­er at 4TU.ResearchData), who also intro­duced learn­ers to best prac­tices for data orga­ni­za­tion of tab­u­lar data based on the Data Car­pen­try for Ecol­o­gists les­son. You can have a look at Eirini’s slides here.

The intro­duc­tion was fol­lowed by a hands-on ses­sion explor­ing the Fric­tion­less Data frame­work. The Fric­tion­less Data project has devel­oped a full data man­age­ment frame­work for Python to describe, extract, val­i­date, and trans­form tab­u­lar data fol­low­ing the FAIR prin­ci­ples. Lil­ly Win­free used Jupyter Note­book to intro­duce learn­ers to the dif­fer­ent tools, as it helps visu­al­iz­ing the steps of the work­flow. You can access the pre­sen­ta­tion and the note­book (and all the mate­ri­als of the work­shop) used by Lil­ly in this GitHub repos­i­to­ry.

Dur­ing the hands-on cod­ing ses­sion, the learn­ers prac­ticed what they were learn­ing on an exam­ple dataset from ecol­o­gy (source of the dataset: Data Car­pen­try for Ecol­o­gists). Lat­er in the work­shop, Kate­ri­na Drak­oula­ki, Fric­tion­less Data fel­low and helper, also gave an exam­ple of how to apply the frame­work tools to a dataset com­ing from the com­pu­ta­tion­al musi­col­o­gy field.

We con­clud­ed the work­shop with a pre­sen­ta­tion about Data Pub­li­ca­tion by Paula Mar­tinez Lavanchy, Research Data Offi­cer at 4TU.ResearchData. The pre­sen­ta­tion focused on why researchers should pub­lish their data, how to select the data to pub­lish and how to choose a good data repos­i­to­ry that helps imple­ment the FAIR prin­ci­ples to the researchers’ data. Paula also briefly demoed the fea­tures of 4TU.ResearchData using the repos­i­to­ry sand­box.

Besides the instruc­tors, we also had a great team of helpers that were there in case the learn­ers encoun­tered any tech­ni­cal prob­lems or had ques­tions dur­ing the live cod­ing ses­sion. We would like to give a big thank you to: Nico­las Dintzn­er – TU Delft Data Stew­ard of the Fac­ul­ty of Tech­nol­o­gy, Pol­i­cy & Man­age­ment, Kate­ri­na Drak­oula­ki – Post­doc­tor­al researcher, at NKUA & Fric­tion­less Data Fel­low, Alek­san­dra Wilczyn­s­ka – Data Man­ag­er at TU Delft Library & the Dig­i­tal Com­pe­tence Cen­ter and Sara Pet­ti – Project Man­ag­er at Open Knowl­edge Foun­da­tion.

Nine­teen learn­ers joined the work­shop. The audi­ence had a broad range of back­grounds with both researchers and sup­port staff (e.g. data cura­tor, research data man­ag­er, research soft­ware engi­neer, data librar­i­an, etc.) rep­re­sent­ed. The work­shop received quite pos­i­tive feed­back. Most of the learner’s expec­ta­tions were ful­filled (79%) and they would rec­om­mend the work­shop to oth­er researchers (93%). It was also nice to know that most of the learn­ers felt that they can apply what they learned imme­di­ate­ly and they felt com­fort­able learn­ing in the work­shop.

Images: Feed­back train­ing event

This feed­back from the learn­ers has helped us to start think­ing about how to improve future runs of the work­shop. For exam­ple, we used less time than we had planned, which cre­ates the oppor­tu­ni­ty to pro­vide instruc­tion on more fea­tures of the frame­work or to add more exer­cis­es or prac­tice time. The learn­ers also indi­cat­ed they would have liked to have a com­mon doc­u­ment (e.g. Google doc or Hack­MD) to share ref­er­ence mate­r­i­al and to doc­u­ment the code that the instruc­tor was typ­ing in case they got lost.

Even though there is room for improve­ment, the learn­ers appre­ci­at­ed  the high­ly prac­ti­cal approach of the work­shop, the space they had to prac­tice what they learned and the over­all qual­i­ty of the Fric­tion­less Data frame­work tools. Here are some of the strengths that learn­ers men­tioned:

‘Hands-on, can start using what I learned imme­di­ate­ly’ 

‘Prac­ti­cal expe­ri­ence with the frame­work and work­ing on shared exam­ples.’ 

‘Machine read­able data and pack­ag­ing for inter­op­er­abil­i­ty through fric­tion­less’ 

‘Very clear con­tent. Assured assis­tance in case of tech­ni­cal prob­lems. Adher­ence to time­lines with breaks. Pro­vid­ed many in-depth links. Friend­ly atmos­phere.’ 

We at the 4TU.ResearchData team great­ly enjoyed this col­lab­o­ra­tion that allowed us to help build the skills that researchers and oth­er users of the repos­i­to­ry need to make research data find­able, acces­si­ble, inter­op­er­a­ble and repro­ducible (FAIR). 

Author: Paula Mar­tinez Lavanchy