Predicción electoral usando un modelo híbrido basado en análisis sentimental y seguimiento a encuestas: elecciones presidenciales de Colombia
Autores/as
-
Mauro Callejas Cuervo
Universidad Pedagógica y Tecnológica de Colombia
https://orcid.org/0000-0001-9894-8737
-
Manuel Andrés Vélez Guerrero
Universidad Pedagógica y Tecnológica de Colombia
https://orcid.org/0000-0002-2105-1742
DOI:
https://doi.org/10.33571/rpolitec.v15n30a9Palabras clave:
PLN, análisis sentimental, resultados electorales, inteligencia artificial, predicciónResumen
La disponibilidad de los medios digitales ha proporcionado una poderosa herramienta para expresar opiniones incluyendo aspectos sociales y políticos desarrollados en cada región. En Colombia, el uso de redes sociales ha dado lugar a la difusión masiva de opiniones políticas, especialmente durante el período de campaña en las elecciones presidenciales nacionales. Este trabajo propone un modelo híbrido para predecir el desenlace de la primera vuelta en las elecciones presidenciales de Colombia en 2018 (pre-hoc), cuyo objetivo es minimizar el error absoluto y mejorar la calidad de la predicción final. Las actividades de los usuarios en Twitter y Facebook fueron registradas y analizadas, obteniendo como resultado una predicción precisa y coherente con la realidad, donde el RMSE del modelo híbrido ronda el 2,47%, superando en promedio el RMSE de las firmas encuestadoras tradicionales más prominentes del país. Adicionalmente también se predijo el valor del abstencionismo electoral con un error diferencial de 1,72% con respecto al valor real, demostrando la confiabilidad de la metodología propuesta.
In Colombia, social networks have become a powerful tool to disseminate political opinions, especially during the campaign period in the national presidential elections. This paper proposes a hybrid model to predict the outcome of the first round of presidential elections in Colombia in 2018, which aims to minimize absolute error and improve the quality of the final prediction. User activities on Twitter and Facebook were recorded and analyzed with artificial intelligence algorithms, resulting in an accurate prediction consistent with reality. As a core result is highlighted that the RMSE of the hybrid model is around 2.47%, surpassing on average the RMSE of the country's most prominent traditional polling firms. Additionally, the value of electoral abstentionism was also predicted with a differential error of 1.72% in relation to the real value, demonstrating the reliability of the proposed methodology.
Métricas de artículo
Resumen: 774 HTML: 973 PDF: 408 XML: 37Métricas PlumX
Citas
K.-B. Shibu Kumar, V.-S. Devi, K. K. Rajeev, y A. Bhatia. Probabilistic algorithms for election result prediction. Proc. - 2014 Int. Conf. Soft Comput. Mach. Intell. ISCMI 2014, pp. 79–82, 2014.
C. Ganser y P. Riordan. Vote expectations at the next level. Trying to predict vote shares in the 2013 German federal election by polling expectations. Elect. Stud., vol. 40, pp. 115–126, 2015.
A. Tumasjan, T. O. Sprenger, P. G. Sandner, y I. M. Welpe. Predicting Elections with Twitter: What 140 Characters Reveal about Political Sentiment. Proc. Fourth Int. AAAI Conf. Weblogs Soc. Media Predict., vol. 30, no. 2, pp. 178–185, 2010.
E. Sang y J. Bos. Predicting the 2011 Dutch Senate Election Results with Twitter. Proc. 13th Conf. Eur. Chapter Assoc. Comput. Linguist., no. 53, pp. 53–60, 2012.
J. Lee y Y. Choi. Expanding affective intelligence theory through social viewing: Focusing on the South Korea’s 2017 presidential election. Comput. Human Behav., vol. 83, pp. 119–128, 2018.
M. Korakakis, E. Spyrou, y P. Mylonas. A survey on political event analysis in Twitter. Proc. - 12th Int. Work. Semant. Soc. Media Adapt. Pers. SMAP 2017, pp. 14–19, 2017.
Y. Arslan, A. Birturk, B. Djumabaev, y D. Küçük. Real-time Lexicon-based sentiment analysis experiments on Twitter with a mild (more information, less data) approach. Proc. - 2017 IEEE Int. Conf. Big Data, Big Data 2017, vol. 2018–Janua, pp. 1892–1897, 2018.
B. Heredia, J. Prusa, y T. Khoshgoftaar. Exploring the Effectiveness of Twitter at Polling the United States 2016 Presidential Election. Proc. - 2017 IEEE 3rd Int. Conf. Collab. Internet Comput. CIC 2017, vol. 2017–Janua, pp. 283–290, 2017.
A. M. Koli, M. Ahmed, y J. Manhas. An Empirical Study on Potential and Risks of Twitter Data for Predicting Election Outcomes. in Emerging Trends in Expert Applications and Security, vol. 841, no. January, Springer Singapore, 2019, pp. 725–731.
Y. Zhao y E. Santos. A Failure of Collective Intelligence. 2018 IEEE/WIC/ACM Int. Conf. Web Intell., pp. 361–366, 2018.
L. Wang y J. Q. Gan. Prediction of the 2017 French election based on Twitter data analysis. 2017 9th Comput. Sci. Electron. Eng. Conf. CEEC 2017 - Proc., pp. 89–93, 2017.
A. Saifuddin, J. Kokil, y M. S. Marko. Tweets & Votes - A 4 Country Comparison of Volumetric and Sentiment Analysis Approcahes. Proc. 10th Int. Conf. Web Soc. Media, no. Icwsm, pp. 507–510, 2016.
M. Ramzan, S. Mehta, y E. Annapoorna. Are tweets the real estimators of election results?. 2017 10th Int. Conf. Contemp. Comput. IC3 2017, vol. 2018–Janua, no. August, pp. 1–4, 2018.
P. Sharma y T. S. Moh. Prediction of Indian election using sentiment analysis on Hindi Twitter. Proc. - 2016 IEEE Int. Conf. Big Data, Big Data 2016, pp. 1966–1971, 2016.
J. Ramteke, S. Shah, D. Godhia, y A. Shaikh. Election result prediction using Twitter sentiment analysis. Proc. Int. Conf. Inven. Comput. Technol. ICICT 2016, vol. 1, 2017.
P. Juneja. Casting Online Votes: To Predict Offline Results Using Sentiment Analysis by machine learning Classifiers. 8th ICCCNT 2017, 2017.
M. Coletto, C. Lucchese, S. Orlando, y R. Perego. Electoral Predictions with Twitter: a Machine-Learning approach Introduction and Related Work. Proc. 6th Ital. Inf. Retr. Work., 2017.
D. Leiter, A. Murr, E. Rascón Ramírez, y M. Stegmaier. Social networks and citizen election forecasting: The more friends the better. Int. J. Forecast., vol. 34, no. 2, pp. 235–248, 2018.
J.A. Caetano, J. Almeida, y H.T. Marques-Neto. Characterizing politically engaged users’ behavior during the 2016 us presidential campaign. IEEE/ACM Int. Conf. Adv. Soc. Networks Anal. Mining, pp. 523–530, 2018.
B. Kostadinov. Predicting the Next US President by Simulating the Electoral College. J. Humanist. Math., vol. 8, no. 1, pp. 64–93, 2018.
A. Hernandez-Suarez, et al. Predicting political mood tendencies based on Twitter data. Proc. - 2017 5th Int. Work. Biometrics Forensics, IWBF 2017, pp. 1–6, 2017.
M.S. Lewis-Beck y C. Tien. Candidates and campaigns: How they alter election forecasts. Elect. Stud., vol. 54, no. March 2017, pp. 303–308, 2018.
S. Martin-Gutierrez, J. C. Losada, y R. M. Benito. Recurrent Patterns of User Behavior in Different Electoral Campaigns: A Twitter Analysis of the Spanish General Elections of 2015 and 2016. Complexity, vol. 2018, pp. 1–15, Dec. 2018.
R. Johnston, T. Hartman, y C. Pattie. Predicting general election outcomes: campaigns and changing voter knowledge at the 2017 general election in England. Quality and Quantity, Springer Netherlands, pp. 1–21, 20-Oct-2018.
D. Hussein. A survey on sentiment analysis challenges. J. King Saud Univ. - Eng. Sci., vol. 30, no. 4, pp. 330–338, 2018.
U. Khan y R.P. Lieli. Information flow between prediction markets, polls and media: Evidence from the 2008 presidential primaries. Int. J. Forecast., vol. 34, no. 4, pp. 696–710, 2018.
M.H. Wang y C.L. Lei. Boosting election prediction accuracy by crowd wisdom on social forums. 2016 13th IEEE Annu. Consum. Commun. Netw. Conf. CCNC 2016, pp. 348–353, 2016.
A. Mavragani y K. P. Tsagarakis. Predicting referendum results in the Big Data Era. J. Big Data, vol. 6, no. 1, p. 3, Dec. 2019.
M. Ankit y N. Saleena. An Ensemble Classification System for Twitter Sentiment Analysis. Procedia Comput. Sci., vol. 132, no. Iccids, pp. 937–946, 2018.
B. Bansal y S. Srivastava. On predicting elections with hybrid topic-based sentiment analysis of tweets. 3rd Int. Conf. Comput. Sci. Comput. Intell. 2018, vol. 135, no. 1, pp. 346–353, 2018.
J. A. Cerón-Guzmán y E. León-Guzmán. A sentiment analysis system of Spanish tweets and its application in Colombia 2014 presidential election. IEEE Int. Conf. Big Data Cloud Comput. BDCloud, pp. 250–257, 2016.
R. Castro y C. Vaca. National leaders’ twitter speech to infer political leaning and election results in 2015 Venezuelan parliamentary elections. IEEE Int. Conf. Data Min. Work, pp. 866–871, 2017.
O. Hidalgo, R. Jaimes, E. Gomez, y S. Lujan-Mora. Sentiment analysis applied to the popularity level of the ecuadorian political leader Rafael Correa. Int. Conf. Inf. Syst. Comput. Sci, vol. 2017–Novem, pp. 340–346, 2018.
S. Rodríguez et al. Forecasting the Chilean electoral year: Using twitter to predict the presidential elections of 2017. in Lecture Notes in Computer Science, vol. 10914 LNCS, pp. 298–314, 2018.
G. Roland, et al. Colombia’s electoral and party system: Proposals for reforms, 2000.
D.M. Hanratty, S.W. Meditz, y R.A. Hudson, Colombia: a country study, vol. 1, no. 1. 2010.
Misión de Observación Electoral MOE. Political Context of the 2018 presidential election in Colombia. Bogotá, 2018.
H. K. Sonneland y Americas Society Council of the Americas. Poll Tracker: Colombia’s 2018 Presidential Election. 2018. Disponible en: https://www.as-coa.org/articles/poll-tracker-colombias-2018-presidential-election. [Consultado el: 25-Feb-2019].
A.P. Torres Espinosa y J. Ferri Durá, Abstención electoral en Colombia. Desafección política, violencia política y conflicto armado, 1st ed. Bogotá: Universidad Complutense de Madrid, 2013.
Registraduría Nacional del Estado Civil de Colombia. Resultados Elecciones Presidenciales 2018 Primera Vuelta. 2018. Disponible en: https://www.colombia.com/elecciones/2018/resultados/presidente.aspx?C=P1. [Consultado el: 27-Feb-2019].
S. J. Taylor and B. Letham. Forecasting at Scale. PeerJ Prepr., pp. 1–25, 2017.
Publicado
Cómo citar
Número
Sección
Licencia
_
