Using Internet Search Data to Predict Well-Being in Japan
Author Information
Author(s): Mavragani Amaryllis, Matsuda Yasumasa, Guo Rui, Boumahdi Ilyes, Yang Myung Si, Taira Kazuya
Primary Institution: Kyoto University
Hypothesis
Can internet search data effectively predict well-being indicators at the regional level in Japan?
Conclusion
The study shows that internet search log data can accurately predict well-being indicators in Japanese prefectures, providing a cost-effective alternative to traditional surveys.
Supporting Evidence
- The best Elastic Net model achieved an R2 of 0.665 when predicting well-being indicators.
- The study utilized Google Trends data to gather search volumes for relevant terms.
- The model was optimized using cross-validation to minimize prediction error.
Takeaway
This study found that looking at what people search for online can help us understand how happy they are in different parts of Japan.
Methodology
The study used Elastic Net methodology to analyze Google Trends data and predict the Regional Well-Being Index (RWI) for Japanese prefectures.
Potential Biases
Potential biases in search data interpretation and the ecological nature of the study limit individual-level insights.
Limitations
The study did not consider causal relationships between search data and well-being indicators and used data from 2019, which may not reflect current trends.
Participant Demographics
The study focused on prefectural data in Japan, without individual-level demographic details.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website