Skip to content
English
  • There are no suggestions because the search field is empty.

Understanding Outliers in the Map Interface

Map Outliers provides for a quick means to see observed changes in the network based on current observations measured against recent averages.

Overview

In the map outliers, there are two types of outliers that can be reviewed. Site based outliers are focused on data that can be reported based on sightings at specific sensors, and Rail Lines based outliers are inferred outliers between sensors, or on likely movements to next junctions in which RailState doesn’t explicitly have sensors. Site based outliers can be suvscribed to, in which you can receive an email notice, and rail line outliers not explicidly focused on sensors can be reviewed on the map.

Further detail on each outlier type, and the categories available are outlined below in this article.

For an additional 'How-To' overview, please see: Map Outliers Overview and Creating a Subscription

SITE BASED OUTLIERS:

Site based outliers are informational indicators used to highlight potential unusual patterns in train traffic as detected by individual RailState sensors. Users can subscribe, to the site-based sensors, and receive email notifications on sensor status and changes detected. These notifications can be a useful first heads-up that changes in the network are being detected.

There are numerous types of Site Based sensor outliers, as listed below with definitions:

  • Sensor Downtime: The sensor has been online less than 95% of the time in the Last 24 hours. (Note: Short downtimes associated with internet connection may not result in any missed train sightings, as sensors continue to collect data and transmit once back online.)
  • No Recent Train: The elapsed time since the last detected train at the sensor site is greater than 200% of the average length of the time interval between two such train detections over the last _______ days.
  • Low Train Count: The number of trains detected by the RailState sensor in the last 24 hours was lower than 50% of the average over the last 30 days.
  • High Train Count: The number of trains detected by the RailState sensor in the last 24 hours was greater than 150% of the average over the last 30 days.
  • Low Train Car Count: The average count of cars of trains detected by the RailState sensor in the last 24 hours was lower than 50% of the average over the last 30 days.
  • High Train Car Count: The average count of cars of trains detected by the RailState sensor in the last 24 hours was greater than 150% of the average over the last 30 days.

 

RAIL LINES BASED OUTLIERS:

Rail section outliers are informational indicators used to highlight potential unusual patterns in train traffic across parts of the rail network where direct sensor data is unavailable. Unlike site based outliers, users are unable to subscribe to these outliers, due to limitations in how rail sections are identified and tracked. We do display them on the map still, however, as they can still be useful to review to see potential changes in patterns.

Why Rail Section Outliers Are Not Subscribable
  • Rail sections are defined as the stretches of rail between junctions (which may be sensors, line ends, or multi-line intersections).
  • Within RailState, these sections are identified by pairs of large, unstable junction IDs (e.g., 2410870003439900), which are not user-friendly or consistent over time.
  • Because of this, we cannot reliably describe or label these sections for users, making subscription impractical – but can still create an interesting visual on the map interface.
How Rail Section Outliers Are Computed
  • The system infers train routes using sensor detections. For each train, it estimates the most likely path taken across the network - although this is not always verifiable based on ‘next’ sensor locations, or areas between RailState sensors.
  • Based on these inferred paths, it compiles a list of trains that passed through or likely entered into each rail section.
  • Metrics (e.g., train count) and outliers (e.g., “No Recent Train”) are then calculated per section using this inferred data.

Example

A train detected at:

  • Welton, AZ
  • Benson, AZ
  • Bovina, TX

…is inferred to have traveled through multiple subdivisions (e.g., Gila, Lordsburg, Deming, El Paso, Clovis, Hereford), even if it wasn’t directly observed at every sensor. The system assumes the train passed through intermediate junctions and sections based on known routes and sensor data.

Limitations

  • These computations rely solely on trains we detect. In areas with low sensor coverage, some trains may go undetected, leading to incomplete or not fully representative metrics.
  • Segment outliers (between adjacent sensors) are more sensitive to sensor downtime, while section outliers can infer data from surrounding sensors, offering broader coverage.

Future Improvements

Although this release of Outliers is an initial iteration, we may eventually assign readable names to rail sections using subdivision names and mileage markers (e.g., “Oregon Trunk from 12.3 mi to 19.7 mi”), but this will require incorporating additional data that is not yet present.

Disclaimer for Users

In the meantime, please note that map metrics and outlier data are based only on rail traffic we detect via sensors. In areas with low sensor coverage, these metrics may not be fully representative due to unseen train movements, but again main still be useful as indicative of what may be happening on the network.

A breakdown of map-based Rail Line Outliers, with definitions, are below:

High Transit Time: The average transit time on the rail segment in the last 24 hours was greater than 150% of the average over the last 30 days.

No Recent Train: The elapsed time since the last detected train traveling on the rail line is greater than 200% of the average length of the time interval between two such train detections over the last ______ days. (This parameter cannot be added to a subscription.)

Low Train Count: The number of detected trains traveling on the rail section in the last 24 hours was lower than 50% of the average over the last 30 days. (This parameter cannot be added to a subscription.)

High Train Count: The number of detected trains traveling on the rail section in the last 24 hours was greater than 150% of the average over the last 30 days. (This parameter cannot be added to a subscription.)

Low Train Car Count: The average count of cars of trains traveling on the rail section in the last 24 hours was lower than 50% of the average over the last 30 days. (This parameter cannot be added to a subscription.)

High Train Car Count: The average count of cars of trains traveling on the rail section in the last 24 hours was greater than 150% of the average over the last 30 days. (This parameter cannot be added to a subscription.)

Reminder: For an additional 'How-To' overview, please see: Map Outliers Overview and Creating a Subscription

 

We will continue to add additional functionality to Outliers over time. In the meantime, please reach out with any additional suggestions.