Web Data Mining: Validity of Data from Google Earth for Food Retail Evaluation.

Area of Science:

• Public Health
• Geospatial Analysis
• Urban Planning

Background:

• Accurate community food retail data is crucial for public health research and interventions.
• Existing methods for data collection can be time-consuming and resource-intensive.
• Geospatial tools offer potential for efficient and large-scale data acquisition.

Purpose of the Study:

• To develop and validate a novel tool for automatic food retail data capture using Google Earth (GE).
• To assess the validity of GE-derived data against ground-truth observations for community food environments.
• To examine if data validity varies by food outlet type and socioeconomic status (SES).

Main Methods:

• A tool was developed to automatically extract food retail data from Google Earth (GE) via a grid point system.
• The study encompassed 300 census tracts stratified by SES in Rio de Janeiro and Belo Horizonte, Brazil.
• GE data was rigorously compared with ground-truth observations to determine validity metrics.

Main Results:

• The GE tool identified 731 food outlets, compared to 856 by ground-truth, demonstrating moderate to excellent validity across all measures (66.3-100%).
• Validity did not significantly differ across socioeconomic status strata.
• Supermarkets, convenience stores, and restaurants showed higher validity compared to other outlet types.

Conclusions:

• Google Earth (GE) presents a viable and innovative tool for automated community food retail data collection.
• The GE-derived data offers satisfactory validity for assessing community food environments across diverse SES areas and outlet types.
• This approach supports scholarly research and non-commercial applications in public health and urban planning.