Schedule

ODISSEI-SICSS is split into two phases. During the first week the participants will be introduced to ODISSEI and taught machine learning in python. This will be primarily done through a course run by the eScience center. They will directly apply this knowledge to worked examples using a consolidated version of the LISS Panel data, with the aim of predicting fertility behavior in the LISS Panel. On Thursday, the participants will be given additional training on analytical methods that they could apply within the benchmarking challenge. On Friday, the participants will work in teams to submit their best prediction model to a benchmarking platform, where their model will be evaluated by the ODISSEI benchmarking team.
At the end of every day during the summer school (except Friday afternoons), there will be a guest speaker who will present their work or a general topic in computational social science to the participants. These talks will last 30-40 mins with 20-30 mins of Q&A.
Slides will be added on the ODISSEI Zenodo Community; other material will be shared via GitHub
All course sessions will take place in the Mandeville Building on the Erasmus Campus Woudestein, Burgemeester Oudlaan 50, Rotterdam. You can find information here and download the campus map here.
Throughout the summer school, ODISSEI will organize various social events. These will be announced closer to the start of the programme.

Session 1: Introducing ODISSEI and Computational Social Science. Speaker: Tom Emery
Session 2: Open Science Workflows in ODISSEI. Speaker: Angelica Maineri
Session 3: Introduction to the theme of the Summer School. Speakers: Elizaveta Sivak & Paulina Pankowska
Programme starts at 10:00, rather than 9:00
Location: Room: T3-13, Mandeville building
Opening Drinks & Bites @ Erasmus Paviljoen, Rotterdam

Session 1 & 2: Introduction to Machine Learning
Session 3: Introduction to LISS
Location: Room: T3-13, Mandeville building
Guest lecture by Jan Kabatek (University of Melbourne): Challenges on working with CBS data and making them accessible

Session 1, 2 & 3: Advanced Machine Learning Using LISS Data
Location: Room: T3-13, Mandeville building
Guest lecture by Gabriele Mari (Erasmus University Rotterdam): Educational Outcomes using CBS Data

Session 1 & 2: Causal Impact Analysis. Speaker: Erik-Jan van Kesteren
Session 3: Intro to Network Analysis. Speaker: Javier Garcia-Bernardo
Location: Room: T3-13, Mandeville building
Guest lecture by Clémentine Cottineau (TU Delft): Agent-Based Modelling of Urban Segregation
Drinks & Dinner at De Gele Kanarie in Rotterdam

In the second week , the teams will be given access to the CBS remote access environment and will be able to link the LISS respondents from week 1 to administrative data and extend their models. They will have one week to work in their teams to improve their prediction models from week 1 using this wider range of predictors with the winners announced on the final Friday of the summer school. For this final element of the challenge, teams will not only be assessed by how well their model predicts the outcomes for the LISS respondents, but also the degree to which it predicts fertility behavior in the wider population (i.e. its generalizability).

SICSS-ODISSEI