A data-driven approach to predicting power outages during winter storms in the southern U.S. leveraging nonparametric machine learning models

Abstract

In February 2021, Winter Storm Uri severely impacted much of the southern United States, triggering unprecedented large-scale power outages. Recognizing that a similar extreme weather event could occur in the future, this study identifies as its primary research objective the development of a baseline power outage prediction model specifically tailored for the southern region of the United States. Central to this objective is the research question: Which variables and which regression models play the most significant role in accurately predicting power outages in this context? Given that large-scale outages are, in essence, a direct result of imbalances between electricity supply and demand, population was considered a key influencing factor. Furthermore, to ensure the model adequately reflects the meteorological characteristics of winter storms, several atmospheric variables—such as dew point and atmospheric pressure—were incorporated into the analysis. These variables are intended to capture the environmental dynamics that underpin outage occurrence during extreme cold events. Four machine learning models—Random Forest, eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost)—were employed in this study. In addition, to enable a comparison between these four machine learning approaches and traditional statistical models, Ridge regression and Lasso regression were also implemented, utilizing population and geographic information data in conjunction with meteorological variables to achieve this objective. To determine the optimal model configuration, Bayesian optimization was employed using tenfold cross-validation. The results revealed that XGBoost achieved the highest performance, with an R² score of 0.92. Furthermore, when the XGBoost model was utilized for prediction, a permutation importance analysis identified population, dew point, and pressure—in that order—as the most influential variables. Additionally, given that the number of data points varied by state during the test evaluation phase, a weighted evaluation metric was also computed using the data counts for each state. Under this weighted evaluation, XGBoost still achieved the highest R² score (0.74), further underscoring its robustness across heterogeneous state-level datasets. Consequently, this paper developed a foundational baseline model for power outage prediction due to winter storms in the Southern United States and identified essential variables through analysis.

Keywords

Power outages / Data-Driven / Winter Storms / Machine learning

References

[1] Allen-Dumas MR, Lee S, Chinthavali S. Analysis of Correlation between Cold Weather Meteorological Variables and Electricity Outages. IEEE International Conference on Big Data (Big Data), 2022, 2022: 3398-3401

[2] Arora P, Ceferino L. Probabilistic and machine learning methods for uncertainty quantification in power outage prediction due to extreme events. Natural Hazards and Earth System Sciences, 2023, 23(5): 1665-1683

[3] Arora P, Ceferino L. A quasi-binomial regression model for hurricane-induced power outages during early warning. ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part a: Civil Engineering, 2024, 10(2): 04024027

[4] Bhattacharyya A, Hastak M. Indirect cost estimation of winter storm-induced power outage in Texas. Journal of Management in Engineering, 2022, 38(604022057

[5] Breiman L. Random forests. Machine Learning, 2001, 45(15-32

[6] Caller-Times Corpus Christi. (2022). OFF THE GRID: United States Power Outage Tracker. Corpus Christi Caller-Times. https://data.caller.com/national-power-outage-map-tracker/

[7] Cerrai D, Koukoula M, Watson P, Anagnostou EN. Outage prediction models for snow and ice storms. Sustainable Energy, Grids and Networks, 2020, 21: 100294

[8] Chen, T., & Guestrin, C. (2016). XGBoost: A Scalable Tree Boosting System. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794. https://doi.org/10.1145/2939672.2939785

[9] Daoud, E. A. (2019, January 15). Comparison between XGBoost, LightGBM and CatBoost Using a Home Credit Dataset. https://www.semanticscholar.org/paper/Comparison-between-XGBoost%2C-LightGBM-and-CatBoost-a-Daoud/b992fdb71b4b78d7b81dc3761402f4eb446077c2

[10] Dorogush, A. V., Ershov, V., & Gulin, A. (2018). CatBoost: Gradient boosting with categorical features support (No. arXiv:1810.11363). arXiv. https://doi.org/10.48550/arXiv.1810.11363

[11] Electric Reliability Council of Texas. (2025). Federal Energy Regulatory Commission. https://www.ferc.gov/industries-data/electric/electric-power-markets/ercot

[12] Friedman JH. Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 2001, 29(5): 1189-1232

[13] Gholamy, A., Kreinovich, V., & Kosheleva, O. (2018). Why 70/30 or 80/20 Relation Between Training and Testing Sets: A Pedagogical Explanation. Departmental Technical Reports (CS). https://scholarworks.utep.edu/cs_techrep/1209

[14] Grineski SE, Collins TW, Chakraborty J. Cascading disasters and mental health inequities: Winter Storm Uri, COVID-19 and post-traumatic stress in Texas. Social Science & Medicine, 2022, 315: 115523

[15] Grineski SE, Collins TW, Chakraborty J, Goodwin E, Aun J, Ramos KD. Social disparities in the duration of power and piped water outages in Texas after Winter Storm Uri. American Journal of Public Health, 2023, 113(130-34

[16] Guikema SD, Quiring SM. Hybrid data mining-regression for infrastructure risk assessment based on zero-inflated data. Reliability Engineering & System Safety, 2012, 99: 178-182

[17] Guikema SD, Davidson RA, Liu H. Statistical models of the effects of tree trimming on power system outages. IEEE Transactions on Power Delivery, 2006, 21(31549-1557

[18] Guikema SD, Quiring SM, Han S-R. Prestorm estimation of hurricane damage to electric power distribution systems. Risk Analysis, 2010, 30(12): 1744-1752

[19] Guikema SD, Nateghi R, Quiring SM, Staid A, Reilly AC, Gao M. Predicting hurricane power outages to support storm response planning. IEEE Access, 2014, 2: 1364-1373

[20] Han S-R, Guikema SD, Quiring SM. Improving the Predictive Accuracy of Hurricane Power Outage Forecasts Using Generalized Additive Models. Risk Analysis, 2009, 29(10): 1443-1453

[21] Han S-R, Guikema SD, Quiring SM, Lee K-H, Rosowsky D, Davidson RA. Estimating the spatial distribution of power outages during hurricanes in the Gulf coast region. Reliability Engineering & System Safety, 2009, 94(2): 199-210

[22] Hoegh-Guldberg, O., Jacob, D., Bindi, M., Brown, S., Camilloni, I., Diedhiou, A., Djalante, R., Ebi, K., Engelbrecht, F., Guiot, J., Hijioka, Y., Mehrotra, S., Payne, A., Seneviratne, S. I., Thomas, A., Warren, R., Zhou, G., Halim, S. A., Achlatis, M., … Zougmoré, R. B. (2018). Impacts of 1.5°C Global Warming on Natural and Human Systems. In V. Masson-Delmotte, P. Zhai, H. O. Pörtner, D. Roberts, J. Skea, P. R. Shukla, A. Pirani, W. Moufouma-Okia, C. Péan, R. Pidcock, S. Connors, J. B. R. Matthews, Y. Chen, X. Zhou, M. I. Gomis, E. Lonnoy, T. Maycock, M. Tignor, & T. Waterfield (Eds.), Global warming of 1.5°C. (pp. 175–311). IPCC Secretariat. https://www.ipcc.ch/sr15/chapter/chapter-3/

[23] Hoerl AE, Kennard RW. Ridge regression: Biased estimation for nonorthogonal problems. Technometrics, 1970, 12(1): 55-67

[24] James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An Introduction to Statistical Learning (Vol. 103). Springer. https://doi.org/10.1007/978-1-4614-7138-7

[25] Jobson JD. Applied Multivariate Data Analysis. Springer, 1991

[26] Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., & Liu, T.-Y. (2017, December 4). LightGBM: A Highly Efficient Gradient Boosting Decision Tree. Neural Information Processing Systems. https://www.semanticscholar.org/paper/LightGBM%3A-A-Highly-Efficient-Gradient-Boosting-Tree-Ke-Meng/497e4b08279d69513e4d2313a7fd9a55dfb73273

[27] Kodaz H, Özşen S, Arslan A, Güneş S. Medical application of information gain based artificial immune recognition system (AIRS): Diagnosis of thyroid disease. Expert Systems with Applications, 2009, 36(2, Part 2): 3086-3092

[28] Kohavi, R. (1995). A study of cross-validation and bootstrap for accuracy estimation and model selection. Proceedings of the 14th International Joint Conference on Artificial Intelligence. 2:1137-1143.

[29] Lee C-C, Maron M, Mostafavi A. Community-scale big data reveals disparate impacts of the Texas winter storm of 2021 and its managed power outage. Humanities and Social Sciences Communications, 2022, 9(1): 335

[30] Lee, S., Choi, J., Jung, G., Tabassum, A., Stenvig, N., & Chinthavali, S. (2023). Predicting Power Outage During Extreme Weather Events with EAGLE-I and NWS Datasets. 2023 IEEE 24th International Conference on Information Reuse and Integration for Data Science (IRI), 211–212. https://doi.org/10.1109/IRI58017.2023.00042

[31] Liang W, Luo S, Zhao G, Wu H. Predicting hard rock pillar stability using GBDT, XGBoost, and LightGBM algorithms. Mathematics, 2020, 8(5765

[32] Lin S, Zhang W, Sheridan S, Mongillo M, DiRienzo S, Stuart NA, Stern EK, Birkhead G, Dong G, Wu S, Chowdhury S, Primeau MJ, Hao Y, Romeiko XX. The immediate effects of winter storms and power outages on multiple health outcomes and the time windows of vulnerability. Environmental Research, 2021, 196 110924

[33] Liu H, Davidson RA, Rosowsky DV, Stedinger JR. Negative binomial regression of electric power outages in hurricanes. Journal of Infrastructure Systems, 2005, 11(4258-267

[34] McRoberts DB, Quiring SM, Guikema SD. Improving hurricane power outage prediction models through the inclusion of local environmental factors. Risk Analysis, 2018, 38(12): 2722-2737

[35] Melaku ND, Fares A, Awal R. Exploring the impact of Winter Storm Uri on power outage, air quality, and water systems in Texas, USA. Sustainability, 2023, 15(5): 4173

[36] Nateghi R, Guikema S, Quiring SM. Power outage estimation for tropical cyclones: Improved accuracy with simpler models. Risk Analysis, 2014, 34(61069-1078

[37] Nejat A, Solitare L, Pettitt E, Mohsenian-Rad H. Equitable community resilience: The case of Winter Storm Uri in Texas. International Journal of Disaster Risk Reduction, 2022, 77: 103070

[38] Popik T, Humphreys R. The 2021 Texas blackouts: Causes, consequences, and cures. Journal of Critical Infrastructure Policy, 2021, 2(1): 47-73

[39] Quiring SM, Zhu L, Guikema SD. Importance of soil and elevation characteristics for modeling hurricane-induced power outages. Natural Hazards, 2011, 58(1365-390

[40] Raschka, S., Liu, Y. (Hayden), Mirjalili, V., & Dzhulgakov, D. (2022). Machine Learning with PyTorch and Scikit-Learn: Develop machine learning and deep learning models with Python. https://ieeexplore.ieee.org/document/10162164

[41] Roustaei N. Application and interpretation of linear-regression analysis. Medical Hypothesis, Discovery and Innovation in Ophthalmology, 2024, 13(3): 151-159

[42] Shashaani S, Guikema SD, Zhai C, Pino JV, Quiring SM. Multi-stage prediction for zero-inflated hurricane induced power outages. IEEE Access, 2018, 6: 62432-62449

[43] Texas Association of Counties. (2022). Texas Association of Counties. https://www.county.org/

[44] The City of Austin and Travis County. (2022). Winter Storm Uri After Action Resources. https://www.austintexas.gov/winter-storm-uri-after-action-resources

[45] Tibshirani R. Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society: Series B (Methodological), 1996, 58(1): 267-288

[46] Tonn GL, Guikema SD, Ferreira CM, Quiring SM. Hurricane Isaac: A longitudinal analysis of storm characteristics and power outage risk. Risk Analysis, 2016, 36(10): 1936-1947

[47] Weather Underground. (2022). Local Weather Forecast, News and Conditions. https://www.wunderground.com/

[48] United States Census Bureau. (2022). United States Census Bureau. Census.Gov. https://www.census.gov/en.html

[49] Wang S, Zhuang J, Zheng J, Fan H, Kong J, Zhan J. Application of Bayesian hyperparameter optimized random forest and XGBoost model for landslide susceptibility mapping. Frontiers in Earth Science, 2021

[50] Wanik DW, Anagnostou EN, Hartman BM, Frediani MEB, Astitha M. Storm outage modeling for an electric distribution network in Northeastern USA. Natural Hazards, 2015, 79(2): 1359-1384

[51] Wanik DW, Parent JR, Anagnostou EN, Hartman BM. Using vegetation management and LiDAR-derived tree height data to improve outage predictions for electric utilities. Electric Power Systems Research, 2017, 146: 236-245

[52] Yadollahie M. The flood in Iran: A consequence of the global warming?. The International Journal of Occupational and Environmental Medicine, 2019, 10(2): 54-56

[53] Yang F, Watson P, Koukoula M, Anagnostou EN. Enhancing weather-related power outage prediction by event severity classification. IEEE Access, 2020, 8: 60029-60042

[54] Zandalinas SI, Fritschi FB, Mittler R. Global warming, climate change, and environmental pollution: Recipe for a multifactorial stress combination disaster. Trends in Plant Science, 2021, 26(6): 588-599

RIGHTS & PERMISSIONS

The Author(s)