Gap filling for Traffic Analytics

The traffic speed observations offered by Traffic Analytics don't cover every road at every minute of every day. This is a natural part of sampling speeds from driven vehicles: if there are no vehicles on a road segment in a specific time period, there are no samples. Even if there are a few vehicles, it's possible that none of them are part of the HERE sample fleet. So, for a given road segment, there may be gaps in the dataset you receive from your query. In some cases, this isn't a problem, either because your analytical process understands and can handle gaps in statistical data, or because you are deriving long-term (or large-area) averages where the chance of completely missing observations is low. However, in other cases, it may be easier to perform analysis when there are no missing values. Traffic Analytics offers a Gap Filling option that ensures there are no gaps in the dataset you receive, by filling in missing observations with substitute speeds. These substitute records are clearly marked by a separate field in the data file (Output Fields).

There are two ways in which to generate a substitute speed: if the gap to fill is 15 minutes or shorter, then the speed supplied is an interpolation between the speeds directly before and after, on the same road segment. In this case, the Confidence field (if selected) is set to the minimum of the two surrounding confidence values, minus five. If the gap to fill is longer than 15 minutes, then the speed supplied is a modeled speed based on historical data for the road segment, taken directly from the HERE Traffic Patterns product. In this case, the Confidence field (if selected) is set to five.

• STDDEV = 0
• MIN = MAX = substitute speed
• PCT-x = substitute speed
• COUNT = 0

This option greatly increases the size of your data file (typically 5X-20X), so use it with care, and only when necessary. Because it generates a substantial amount of data, you should use it in queries whose time duration is one month or less (although it's acceptable to use longer periods if you are only selecting a few individual road segments).

Be aware that gap filling only operates on roads that have at least one observed speed in your query. Road segments which have no observations in a query don't generate records, regardless of whether or not you select gap filling. This means that gap filling doesn't cause Traffic Analytics to generate records for every road segment in the query area; it only covers segments that have at least one probe observation.

Gap filling also operates in one specific kind of gap: when the local time changes from standard time to daylight saving time (“summer” time). In locations where to observe this shift, gap filling adds records through the “gap” even though it didn't represent an actual hour on the ground.