Predicting party switching through machine learning and open data
Nicolò Meneghetti,
Fabio Pacini,
Francesca Biondi Dal Monte,
Marina Cracchiolo,
Emanuele Rossi,
Alberto Mazzoni,
Silvestro Micera
Affiliations
Nicolò Meneghetti
The Biorobotics Institute, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Department of Excellence for Robotics and AI, Scuola Superiore Sant’Anna, 56025 Pisa, Italy
Fabio Pacini
Dirpolis Institute, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Department of Linguistics, Literature, History, Philosophy and Law (DISTU), Tuscia University, 01100 Viterbo, Italy
Francesca Biondi Dal Monte
Dirpolis Institute, Scuola Superiore Sant’Anna, 56025 Pisa, Italy
Marina Cracchiolo
The Biorobotics Institute, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Department of Excellence for Robotics and AI, Scuola Superiore Sant’Anna, 56025 Pisa, Italy
Emanuele Rossi
Dirpolis Institute, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Health Interdisciplinary Center, Scuola Superiore Sant’Anna, 56025 Pisa, Italy
Alberto Mazzoni
The Biorobotics Institute, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Department of Excellence for Robotics and AI, Scuola Superiore Sant’Anna, 56025 Pisa, Italy
Silvestro Micera
The Biorobotics Institute, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Department of Excellence for Robotics and AI, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Health Interdisciplinary Center, Scuola Superiore Sant’Anna, 56025 Pisa, Italy; Bertarelli Foundation Chair in Translational Neural Engineering, Center for Neuroprosthetics and Institute of Bioengineering, Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland; Corresponding author
Summary: Parliament dynamics might seem erratic at times. Predicting future voting patterns could support policy design based on the simulation of voting scenarios. The availability of open data on legislative activities and machine learning tools might enable such prediction. In our paper, we provide evidence for this statement by developing an algorithm able to predict party switching in the Italian Parliament with over 70% accuracy up to two months in advance. The analysis was based on voting data from the XVII (2013–2018) and XVIII (2018–2022) Italian legislature. We found party switchers exhibited higher participation in secret ballots and showed a progressive decrease in coherence with their party’s majority votes up to two months before the actual switch. These results show how machine learning combined with political open data can support predicting and understanding political dynamics.
