Research Article

Transferability of a Machine Learning-Based Model of Hourly Traffic Volume Estimation—Florida and New Hampshire Case Study

Table 1

Ground truth and input variables used for model development.

NameCategoryData typeDescription

Ground truth volumes
CCS traffic countsInteger (≥0)Hourly traffic counts from continuous count stations
Input data features
Probe counts (weight class 1)Probe countsInteger (≥0)Number of unique probe counts from weight class 1 vehicles (<14k lbs)
Probe counts (weight class 2)Integer (≥0)Number of unique probe counts from weight class 2 vehicles (14k–26k lbs)
Probe counts (weight class 3)Integer (≥0)Number of unique probe counts from weight class 3 vehicles (>26k lbs)
Probe count null flagBinary (0, 1)Indicator variable used to identify persistent missing probe data
Avg probe speedProbe speedsFloat (≥ 0)Average (harmonic) probe speed over 1 hour time period (mph)
Reference speedFloat (≥ 0)Estimated freeflow speed for segment (mph)
Avg travel timeFloat (>0)Avg time to travel across segment (seconds)
Probe speed null flagBinary (0, 1)Indicator variable used to identify persistent missing probe speed/travel times
TemperatureWeatherFloat (≥ 0)Temperature in degrees Fahrenheit
Temperature null flagBinary (0, 1)Indicator variable used to identify whether temperature is missing
DewpointFloat (≥ 0)Dew point in degrees Fahrenheit
Dewpoint null flagBinary (0, 1)Indicator variable used to identify whether dewpoint is missing
Relative humidityFloat (≥ 0)Relative humidity (%)
Relative humidity null flagBinary (0, 1)Indicator variable used to identify whether relative humidity is missing
VisibilityFloat (≥ 0)Visibility (miles)
Visibility null flagBinary (0, 1)Indicator variable used to identify whether visibility is missing
PrecipitationFloat (≥ 0)Precipitation (inches)
Precipitation null flagBinary (0, 1)Indicator variable used to identify whether precipitation is missing
Segment lengthRoad characteristicsFloat (≥0)Length of segment in miles
OSM road classBinary (0, 1)6 one-hot encoded variables representing possible OpenStreetMap road classes
OSM lanesFloat (≥0)OSM-based estimate of number of directional lanes on the road
OSM lanes null flagBinary (0, 1)Indicator variable used to identify whether OSM lanes is missing
Structure typeCategoricalIndicates presence of a special facility (bridge, tunnel, causeway)
HPMS through lanesFloat (≥0)HPMS-based estimate of number of total lanes on the road
Functional system (HPMS)Binary (0, 1)3 one-hot encoded variables for HPMS functional classes 1, 2, 3+
Functional road class (TMC)Binary (0, 1)3 one-hot encoded variables for TMC functional classes 1, 2, 3+
AADTFloat (≥0)HPMS annual avg daily traffic associated with TMC segment
AADT (single)Float (≥0)Single-unit truck and bus AADT associated with TMC segment
AADT (combination)Float (≥0)Combination truck AADT associated with TMC segment
HourTemporalBinary (0, 1)24 one-hot encoded variables representing hours 0–23
WeekendBinary (0, 1)2 one-hot encoded variables for identifying Saturday and Sunday
Volume profile estimateOtherFloat (≥0)Time-of-day volume estimate based on TTI profiling method
Volume profile null flagBinary (0, 1)Indicator variable used to identify whether volume profile estimate is missing