# BigQuery-Geotab Intersection Congestion — Data

来源：https://www.kaggle.com/competitions/bigquery-geotab-intersection-congestion/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

> UPDATE: This dataset is no longer available via BigQuery. However, you can source it directly from the Kaggle dataset on this competition.


The data consists of aggregated trip logging metrics from commercial vehicles, such as semi-trucks. The data have been grouped by  intersection, month, hour of day, direction driven through the intersection, and whether the day was on a weekend or not.

For each grouping in the test set, you need to make predictions for three different quantiles of two different metrics covering how long it took the group of vehicles to drive through the intersection. Specifically, the 20th, 50th, and 80th percentiles for the total time stopped at an intersection and the distance between the intersection and the first place a vehicle stopped while waiting. You can think of your goal as summarizing the distribution of wait times and stop distances at each intersection. 

Each of those six predictions goes on a new row in the submission file. Read the submission `TargetId` fields, such as `1_1`, as the first number being the `RowId` and the second being the metric id. You can unpack the submission metric id codes with `submission_metric_map.json`.

The training set includes an optional additional output metric (`TimeFromFirstStop`) in case you find that useful for building your models. It was only excluded from the test set to limit the number of predictions that must be made.

The instructions in `BigQuery-Dataset-Access.md` are now deprecated. The dataset is only available through Kaggle, via download below or API.