Scholarlink

<p>The decline of local newspapers has led to the emergence of news deserts—areas lacking access to critical local information—posing a threat to community engagement and democracy. This study aims to predict which U.S. counties are most at risk of becoming news deserts by developing machine learning models based on socioeconomic, geographic, and circulation data. Addressing class imbalance and data noise, we employed classifiers such as Logistic Regression, Random Forest, XGBoost, Support Vector Machines, K-Nearest Neighbors, and Naive Bayes, combined with resampling techniques like SMOTE, Tomek Links, SMOTETomek, SMOTEENN, and ADASYN. Our analysis found that XGBoost combined with ADASYN performed best, achieving an F2-Score of 0.486 and AUC-PR of 0.467 on test data. These results provide valuable insights for policymakers aiming to develop targeted interventions to preserve local media ecosystems and strengthen democratic processes.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predicting news deserts using supervised machine learning

  • 期刊论文

  • 发表日期：

    2025年3月4日

查看全文
  • Arijit Paladhi

摘要

The decline of local newspapers has led to the emergence of news deserts—areas lacking access to critical local information—posing a threat to community engagement and democracy. This study aims to predict which U.S. counties are most at risk of becoming news deserts by developing machine learning models based on socioeconomic, geographic, and circulation data. Addressing class imbalance and data noise, we employed classifiers such as Logistic Regression, Random Forest, XGBoost, Support Vector Machines, K-Nearest Neighbors, and Naive Bayes, combined with resampling techniques like SMOTE, Tomek Links, SMOTETomek, SMOTEENN, and ADASYN. Our analysis found that XGBoost combined with ADASYN performed best, achieving an F2-Score of 0.486 and AUC-PR of 0.467 on test data. These results provide valuable insights for policymakers aiming to develop targeted interventions to preserve local media ecosystems and strengthen democratic processes.