Dynamic logistic regression and variable selection: Forecasting and contextualizing civil unrest

Bakerman, Jordan; Pazdernik, Karl; Korkmaz, Gizem; Wilson, Alyson G.

doi:10.1016/j.ijforecast.2021.07.003

Dynamic logistic regression and variable selection: Forecasting and contextualizing civil unrest

Journal Article · Wed Feb 16 00:00:00 EST 2022 · International Journal of Forecasting

DOI:https://doi.org/10.1016/j.ijforecast.2021.07.003· OSTI ID:1853807

Bakerman, Jordan ^[1]; ^[2]; ^[3]; ^[1]

North Carolina State Univ., Raleigh, NC (United States)
North Carolina State Univ., Raleigh, NC (United States); Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
Univ. of Virginia, Charlottesville, VA (United States)

Civil unrest can range from peaceful protest to violent furor, and researchers are working to monitor, forecast, and assess such events to allocate resources better. Twitter has become a real-time data source for forecasting civil unrest because millions of people use the platform as a social outlet. Additionally, daily word counts are used as model features, and predictive terms contextualize the reasons for the protest. To forecast civil unrest and infer the reasons for the protest, we consider the problem of Bayesian variable selection for the dynamic logistic regression model and propose using penalized credible regions to select parameters of the updated state vector. This method avoids the need for shrinkage priors, is scalable to high-dimensional dynamic data, and allows the importance of variables to vary in time as new information becomes available. A substantial improvement in both precision and F1-score using this approach is demonstrated through simulation. Finally, we apply the proposed model fitting and variable selection methodology to the problem of forecasting civil unrest in Latin America. Our dynamic logistic regression approach shows improved accuracy compared to the static approach currently used in event prediction and feature selection.

View Accepted Manuscript (DOE)

Research Organization:: Pacific Northwest National Laboratory (PNNL), Richland, WA (United States)

Sponsoring Organization:: National Science Foundation (NSF); USDOE Laboratory Directed Research and Development (LDRD) Program; USDOE National Nuclear Security Administration (NNSA), Office of Defense Nuclear Nonproliferation

Grant/Contract Number:: AC05-76RL01830; NA0002576

OSTI ID:: 1853807

Report Number(s):: PNNL-SA-135434

Journal Information:: International Journal of Forecasting, Journal Name: International Journal of Forecasting Journal Issue: 2 Vol. 38; ISSN 0169-2070

Publisher:: ElsevierCopyright Statement

Country of Publication:: United States

Language:: English

References (24)

Predicting crime using Twitter and kernel density estimation Gerber, Matthew S. Decision Support Systems, Vol. 61 https://doi.org/10.1016/j.dss.2014.02.003	journal	May 2014
Dynamic model averaging in large model spaces using dynamic Occam׳s window Onorante, Luca; Raftery, Adrian E. European Economic Review, Vol. 81 https://doi.org/10.1016/j.euroecorev.2015.07.013	journal	January 2016
Time-varying sparsity in dynamic regression models Kalli, Maria; Griffin, Jim E. Journal of Econometrics, Vol. 178, Issue 2 https://doi.org/10.1016/j.jeconom.2013.10.012	journal	February 2014
Using dynamic model averaging in state space representation with dynamic Occam’s window and applications to the stock and gold market Risse, Marian; Ohl, Ludwig Journal of Empirical Finance, Vol. 44 https://doi.org/10.1016/j.jempfin.2017.09.005	journal	December 2017
Twitter as a tool for forecasting stock market movements: A short-window event study Nisar, Tahir M.; Yeung, Man The Journal of Finance and Data Science, Vol. 4, Issue 2 https://doi.org/10.1016/j.jfds.2017.11.002	journal	June 2018
Twitter mood predicts the stock market Bollen, Johan; Mao, Huina; Zeng, Xiaojun Journal of Computational Science, Vol. 2, Issue 1 https://doi.org/10.1016/j.jocs.2010.12.007	journal	March 2011
Social media data analytics to improve supply chain management in food industries Singh, Akshit; Shukla, Nagesh; Mishra, Nishikant Transportation Research Part E: Logistics and Transportation Review, Vol. 114 https://doi.org/10.1016/j.tre.2017.05.008	journal	June 2018
Dynamic Generalized Linear Models and Bayesian Forecasting West, Mike; Harrison, P. Jeff; Migon, Helio S. Journal of the American Statistical Association, Vol. 80, Issue 389 https://doi.org/10.1080/01621459.1985.10477131	journal	March 1985
Variable Selection via Gibbs Sampling George, Edward I.; McCulloch, Robert E. Journal of the American Statistical Association, Vol. 88, Issue 423 https://doi.org/10.1080/01621459.1993.10476353	journal	September 1993
Consistent High-Dimensional Bayesian Variable Selection via Penalized Credible Regions Bondell, Howard D.; Reich, Brian J. Journal of the American Statistical Association, Vol. 107, Issue 500 https://doi.org/10.1080/01621459.2012.716344	journal	August 2012
Bayesian Inference for Logistic Models Using Pólya–Gamma Latent Variables Polson, Nicholas G.; Scott, James G.; Windle, Jesse Journal of the American Statistical Association, Vol. 108, Issue 504 https://doi.org/10.1080/01621459.2013.829001	journal	December 2013
Bayesian Analysis of Latent Threshold Dynamic Models Nakajima, Jouchi; West, Mike Journal of Business & Economic Statistics, Vol. 31, Issue 2 https://doi.org/10.1080/07350015.2012.747847	journal	April 2013
On Gibbs sampling for state space models Carter, C. K.; Kohn, R. Biometrika, Vol. 81, Issue 3 https://doi.org/10.1093/biomet/81.3.541	journal	January 1994
Markov chain Monte Carlo for dynamic generalised linear models Gamerman, D. Biometrika, Vol. 85, Issue 1 https://doi.org/10.1093/biomet/85.1.215	journal	March 1998
Non-Gaussian Ornstein-Uhlenbeck-based models and some of their uses in financial economics Barndorff-Nielsen, Ole E.; Shephard, Neil Journal of the Royal Statistical Society: Series B (Statistical Methodology), Vol. 63, Issue 2 https://doi.org/10.1111/1467-9868.00282	journal	May 2001
Forecasting Inflation Using Dynamic Model Averaging* Koop, Gary; Korobilis, Dimitris International Economic Review, Vol. 53, Issue 3 https://doi.org/10.1111/j.1468-2354.2012.00704.x	journal	July 2012
Dynamic Logistic Regression and Dynamic Model Averaging for Binary Classification McCormick, Tyler H.; Raftery, Adrian E.; Madigan, David Biometrics, Vol. 68, Issue 1 https://doi.org/10.1111/j.1541-0420.2011.01645.x	journal	August 2011
An Overview of Dynamic Model Averaging Techniques in Time‐Series Econometrics Nonejad, Nima Journal of Economic Surveys, Vol. 35, Issue 2 https://doi.org/10.1111/joes.12410	journal	January 2021
Activism via attention: interpretable spatiotemporal learning to forecast protest activities Ertugrul, Ali Mert; Lin, Yu-Ru; Chung, Wen-Ting EPJ Data Science, Vol. 8, Issue 1 https://doi.org/10.1140/epjds/s13688-019-0183-y	journal	February 2019
The Bayesian Lasso Park, Trevor; Casella, George Journal of the American Statistical Association, Vol. 103, Issue 482 https://doi.org/10.1198/016214508000000337	journal	June 2008
Online Prediction Under Model Uncertainty via Dynamic Model Averaging: Application to a Cold Rolling Mill Raftery, Adrian E.; Kárný, Miroslav; Ettler, Pavel Technometrics, Vol. 52, Issue 1 https://doi.org/10.1198/TECH.2009.08104	journal	February 2010
Least angle regression Tibshirani, Robert; Johnstone, Iain; Hastie, Trevor The Annals of Statistics, Vol. 32, Issue 2 https://doi.org/10.1214/009053604000000067	journal	April 2004
The Digital Evolution of Occupy Wall Street Conover, Michael D.; Ferrara, Emilio; Menczer, Filippo PLoS ONE, Vol. 8, Issue 5 https://doi.org/10.1371/journal.pone.0064679	journal	May 2013
Dynamic Model Averaging in Economics and Finance with fDMA: A Package for R Drachal, Krzysztof Signals, Vol. 1, Issue 1 https://doi.org/10.3390/signals1010004	journal	July 2020

Similar Records

Logistic Regression in Clinical Studies

Journal Article · Mon Feb 14 23:00:00 EST 2022 · International Journal of Radiation Oncology, Biology and Physics · OSTI ID:23200907

Evaluating penalized logistic regression models to predict Heat-Related Electric grid stress days

Journal Article · Wed Nov 01 00:00:00 EDT 2017 · Applied Energy · OSTI ID:1406671

Evaluating penalized logistic regression models to predict Heat-Related Electric grid stress days

Journal Article · Thu Sep 21 20:00:00 EDT 2017 · Applied Energy · OSTI ID:1395358

Related Subjects

97 MATHEMATICS AND COMPUTING
Pólya-Gamma latent variable
civil unrest
dynamic logistic regression
forecasting
penalized credible regions

Dynamic logistic regression and variable selection: Forecasting and contextualizing civil unrest

Citation Formats

References (24)

Similar Records

Related Subjects