Is more spatial data always better? Is there a point where the quality and details in a dataset can be outweighed by an extremely large amount of data lacking any detail beyond time stamps and coordinates?
The article I chose this week addresses big data and the issue of
quantity vs. quality. The study focused on comparing the accuracy
identifying individual and household travel activity and behavior
between traditional data collection surveys and GPS based surveys.
Traditionally this data is collected through surveys where people logged
their activity for a one or two day period and sent in their survey
response. The information includes where and when people traveled as
well as mode of travel and purpose of travel. Voluntary participation
and survey costs limit the data that can be collected. GPS based
surveys, on the other hand, can easily collect data from hundreds of
thousands of people with the acceptance of an application on a smart
phone. The time and travel as easily collected with no additional user
input but this large amount of data lacks details about method of travel
and purpose of travel. This is where Python and spatial analysis come
in.
The study used conducted three experiments analyzing GPS based travel
surveys on one study using traditional survey data from San Francisco.
Each experiment used the same data but a different algorithm to process
the data. The San Francisco data was ran through the algorithms as well
to compare the results for accuracy. The purpose of the analysis was to
try to correctly discern the method of travel and purpose of travel
using spatial analysis as well as extra data from accelerometers and
Wi-Fi devices contained in phones. Scripts were written using SciPY
functions and were applied to the spatial data.
The results showed that big data was useful in identifying patterns
in where and when people travel but was not accurate in determining the
method of travel or purpose of travel. The smaller the sample size the
more accurate the algorithms were but this is the opposite of what is
necessary to process big data. If details aren’t necessary for the
travel study then big data is a great option, but if purpose and method
of travel are needed, study participants will still need to provide
information beyond simply carrying their phones.
Title and Link - When is big data big enough? Implications of using GPS-based surveys for travel demand analysis
Or the DOI: doi:10.1016/j.trc.2015.04.025
No comments:
Post a Comment