Reinterpreting the happiness index using artificial neural networks with selected economic and environmental variables: evidence from D-8 countries


Atalay A.

CURRENT PSYCHOLOGY, cilt.44, sa.12, ss.12760-12770, 2025 (SSCI) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 44 Sayı: 12
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1007/s12144-025-07996-5
  • Dergi Adı: CURRENT PSYCHOLOGY
  • Derginin Tarandığı İndeksler: Social Sciences Citation Index (SSCI), Scopus, IBZ Online, BIOSIS, Business Source Elite, Business Source Premier, Psycinfo
  • Sayfa Sayıları: ss.12760-12770
  • Anahtar Kelimeler: Artificial neural networks (ANN), D8 countries, Happiness, Levenberg–Marquardt method, World happiness report
  • Atatürk Üniversitesi Adresli: Evet

Özet

In recent years, the lifestyle focused on growth, development, and sustainability has motivated numerous national and international organizations. The development level of countries is now measured not only by economic indicators but also by the happiness levels of individuals, making the concept of 'happiness' a significant topic in economic studies. Although happiness was traditionally considered a subfield of psychology and philosophy, it is now also addressed in the field of economics. To maximize happiness, it is necessary to examine the understanding of happiness at monetary, individual, family, community, and environmental (nature) levels. In this study, using the World Happiness Report (WHR) scores of countries between 2012 and 2021, the happiness rates for the D8 countries (Indonesia, Bangladesh, Iran, Egypt, Malaysia, Pakistan, Nigeria, and Turkey), which are developing nations, were estimated using the Artificial Neural Network (ANN) method. The data used for these estimations included per capita GDP, inflation rate, carbon dioxide emissions, freedom report, corruption index, and life expectancy data. In the ANN model established in the study, 70% of the database data was randomly allocated for training, 15% for validation, and 15% for testing. This artificial neural network was trained using the Levenberg-Marquardt method. The regression R values, which are performance indicators of the model, were determined to be 0.99 for training data, 0.97 for validation data, and 0.95 for test data. The regression R-value for the overall data used in the model was determined to be 0.98.