Overview
Each lane in your brokerage's historical data has a set of features, such as origin and weight. A Greenscreens machine learning model recognizes similarities and connections between these features, which helps it to make accurate predictions, even for new lanes. The Similar Lanes Widget is a trend line in the network rate prediction graph that shows a summary of rates for lanes with features similar to those of the lane in question.
Machine learning is not usually self-referential. The Greenscreens model can't tell you how it predicted a rate. What we can do is show the data pool of similar lanes, so you can compare and verify the prediction.
This is especially useful when you're looking at a prediction for a lane new to your brokerage. When there is no historical data for a lane, Greenscreens can still provide a reliable prediction by drawing on data from other lanes. Maybe no one has historical data on a shipment from Butte, MT to Eureka Springs, AR., but Greenscreens can still come up with an accurate prediction on the lane, because there’s plenty of data available for shipments from Spokane to Little Rock that use the same corridor.
Using The Similar Lanes Widget
Reading the Graph
The orange line on the Network Rate Prediction graph represents a summary of similar lane rates. This shows you the general trend of rates for lanes similar to the one in the prediction.
The orange field represents the overall range of aggregated data on similar lanes rates. This shows you the span of rates for similar lanes in your own historical data and throughout the Greenscreens network:
The number in green is the number of similar lanes with data available. You can also click this number to bring up the Similar Lanes Table:
You can toggle the Similar Lanes trend line and the Network Rates trend line off and on by clicking Network or Similar below the graph:
When Similar Lanes are toggled on, you'll see numbers for similar lanes in the Details window when you hover over a point on the graph:
Similar Lanes Table
The Similar Lanes Table shows the eight lanes closest to the prediction, along with the number of known loads that fall into each lane. To bring up the table, click the green link that shows the number of Similar Lanes:
The image below shows the Similar Lanes Table for a van load from Albany, NY to Chicago, IL. You can see that the lanes in the table go from New York to various markets in Illinois. These are the eight loads most similar to the Albany - Chicago van route in the prediction.
Criteria for Similarity
Several criteria define similarity. The main ones are:
Similar lanes should be from GSNetwork.
Mileage should differ from the mileage of the target lane by no more than 20%.
Pickup and dropoff points should be close to those of the target lane. and defined by a formula. For each lane which passes the first two filters, we calculate the distance between the pickup location of the original lane and the pickup location of the tested lane, sort the results in ascending order, and select the top lanes.