A Horvitz-type estimation on incomplete traffic accident data analyzed via a zero-inflated Poisson model

Area of Science:

• Traffic Safety Analysis
• Statistical Modeling
• Data Science

Background:

• Road safety policy relies on accurate traffic data analysis.
• Road crash count data frequently exhibit an excess of zero counts.
• Missing data are a common issue in traffic accident datasets, potentially biasing results.

Purpose of the Study:

• To develop a reliable statistical approach for analyzing road crash data with excess zeros and missing values.
• To provide policy makers with trustworthy data analysis for enacting effective traffic policies.

Main Methods:

• Employed the Horvitz method, which utilizes inverse weighting of observed data.
• Weights for the Horvitz method were derived using both parametric and nonparametric approaches.
• Validated the method through Monte Carlo simulations and analysis of real-world traffic accident data.

Main Results:

• The Horvitz method demonstrated satisfactory performance in handling missing data in road crash counts.
• The approach effectively addressed the challenges posed by excess zeros and non-random missing data.
• Reliable estimators were achieved, improving the trustworthiness of the analysis.

Conclusions:

• The Horvitz method provides a robust solution for statistical modeling of road traffic accident data with missing values.
• Accurate data analysis is essential for informed policy-making to enhance road safety.
• This approach offers a reliable alternative to naive estimation methods when dealing with incomplete traffic data.