Data Fusion: HTS with other Data Sources

What is the problem?

As the type and variety of alternative sources of travel behavior data become more prominent and the costs of collecting travel behavior data by conventional means increase, there is significant interest in identifying and exploring analyses of this data in conjunction with traditional household travel survey data. Preliminary studies have shown that it is possible to fuse traditional household diary data with these alternative data sources and point to the potential for the resultant data to inform travel demand models. The fused data is also anticipated to strengthen our understanding of travel behavior for both local and long-distance travel, particularly with respect to variation in travel over both short and long periods of time.

Why is this a problem?

Household travel survey data informs many policy questions and provides travel behavior data to support travel demand model development. The more advanced travel demand models require more detailed and accurate data than the general public is willing to provide, leading to costly survey designs with low response rates and underrepresentation of important portions of the population. Many believe that alternative data sources can be used to expand and enhance the understanding of travel behavior, providing the advanced travel demand models with in-depth data.

Alongside the interest and opportunity of leveraging these two data sources are several challenges that require investigation. Most notably, the potential to incorporate fused data into the travel models is becoming a reality, but the area is still undefined with respect to acceptance criteria by the federal funding sources (namely FTA and FHWA). Another challenge is that most studies have been in smaller scale with a limited application. Therefore, a solid research effort is needed to demonstrate that fused data and supporting imputation efforts can produce information that is reasonably close to the ground truth. This would help agencies that may be interested in exploring and adopting data fusion but have the challenge of justifying the expense of the alternative data sources and educating their decision makers who must approve the investment.

What is the research need?

Two areas of research are needed:

1. A solid synthesis and understanding of what we know and do not know about data fusion specifically with respect to HTS data. What are the alternative data sources available and which are the best “fit” with the diary data? What are the opportunities for new data sources and what are the barriers or risks threatening the quality and sustainability of the currently available data sources? What are the lessons learned from the proof of concept and smaller scale applications? What analyses can benefit from the fused data?
2. Guidance and direction from oversight agencies and oversight boards. What aspects of data fusion need to be proven or explored in order to secure approval for use in models and funding to purchase alternative data sources? What is the value of data fusion and how can we educate decision-makers and make them more comfortable with results based on fused data?