Dos and Don’ts of Data Analytics in Infrastructure Management

Data analytics has become popular in infrastructure and construction management. Since the culture open data and model and data governance is still not well developed in the construction industry, best practices in this field are usually derived from other domains. This paper presents best practices in the application of data analytics to infrastructure management with emphasis on road asset management. The paper investigates four main areas: data retrieval and preparation, data quality, algorithm, and implementation. The paper relies on the pavement performance data retrieved from the Long-Term Pavement Performance (LTPP) with specific focus on two pavement performance indicators: Pavement Condition Index (PCI) and International Roughness Index (IRI). More than 5000 records of PCI and 40,000 IRI records were used to demonstrate the best practices. Using this data, the paper investigated the barriers to implementing data analytic solutions with emphasis on data quality issues (e.g., data representativeness, scarcity, and sanity) and model training (e.g., accuracy and overfitting). Identification of population shift and limitations of algorithms when applied to certain types of data were investigated. The recommendations presented in this paper can be used in different sectors of infrastructure management.