# Chinese Railways delay prediction-public — Data

来源：https://www.kaggle.com/competitions/chinese-railways-delay-prediction-public/data

采集时间：2026-10-20T09:31:52Z

English:​​

This competition challenges participants to predict ​the arrival delay (in minutes) at the final station​ of the high-speed train G77 service ("G77" is the train ID). The training data consists of historical journey records for this specific train, detailing its schedule and performance at each station along its route between Beijing and Guangzhou South. The test data provides the scheduled information for a future journey of the same train service (G77) but omits the delay information at the final station (Guangzhou South). Competitors must build a model using the training data to predict the delay minutes upon arrival at Guangzhou South in the test data.

​中文 (Chinese):​​

本次竞赛要求参赛者预测高铁 G77 次列车（"G77"为列车车次号）​在终点站（广州南）的到达延误时长（单位：分钟）​。训练数据包含该特定列车的历史运行记录，详细描述了其在北京至广州南沿途各车站的时刻表与实际运行情况。测试数据提供了同一列车（G77）未来一趟旅程的时刻表信息，但缺少终点站（广州南）的延误信息。参赛者必须利用训练数据构建模型，以预测测试数据中列车到达广州南时的延误分钟数。

Files / 文件
​train.csv​ - The ​primary training data. Comprised of historical records for the ​G77 train​ service. Each row represents a scheduled stop (station) along a specific journey on a specific date. One file per journey date exists initially (e.g., 20250703_G77.csv), but ​competitors must merge all [YYYYMMDD]_G77.csvfiles into a single train.csvfile​ for model training. Contains the 延误分钟 (Delay Minutes)column.

​主训练数据。包含 ​G77 次列车的历史运行记录。每一行代表特定日期、特定车次在沿途某一车站的时刻信息。初始状态为每个旅程日期一个文件（例如 20250703_G77.csv），但参赛者必须将所有 [YYYYMMDD]_G77.csv文件合并成一个 train.csv文件用于模型训练。包含 延误分钟 (Delay Minutes)列。

​test.csv​ - The ​test set. Contains the ​scheduled information​ for one future journey (dated 2025-10-14) of the ​G77 train​ service from Beijing to Guangzhou South. The 延误分钟 (Delay Minutes)column is ​MISSING​ for the final station (广州南/ Guangzhou South) - this is the target variable to predict.

​测试集。包含 ​G77 次列车​ 一趟未来旅程（日期 2025-10-14）从北京开往广州南的时刻表信息。终点站 (广州南/ Guangzhou South) 的 延误分钟 (Delay Minutes)列缺失——这正是需要预测的目标变量。

​sample_submission.csv​ - A sample submission file demonstrating the required format. It will have one row corresponding to the final station (Guangzhou South/ 广州南) entry in test.csvwith the predicted 延误分钟 (Delay_Minutes).

​样本提交文件。展示所需格式的示例文件。该文件将有一行，对应于 test.csv中终点站 (广州南/ Guangzhou South) 的记录，并包含预测的 延误分钟 (Delay_Minutes)值。

​metaData.csv​ - (Optional/Recommended)A file providing supplemental information, such as:

Station names and potentially latitude/longitude or region.

Explanation of holidays/events during journey dates.

Weather codes/stats for the journey dates/stations. Not provided in the initial sample.

​元数据文件​ - (可选/推荐)提供补充信息的文件，例如：

车站名称以及可能的经纬度或区域信息。

运行日期期间节假日/特殊事件的说明。

运行日期及车站的天气代码/统计数据。初始样本中未提供此文件。

Columns / 数据列
​车次ID (Train_ID): Unique identifier for the train service. ​Always G77in this dataset. Indicates data is solely for this specific train route.

​列车车次标识符。该数据集中始终为 G77。表示所有数据仅针对这一特定列车运行路线。

Format:String (e.g., G77).

Importance:Key identifier. All provided data belongs to this train service.

​车站名 (Station_Name): Name of the train station. The route sequence is 北京(Beijing) -> 北京西(Beijing West) -> 石家庄(Shijiazhuang) -> 郑州东(Zhengzhou East) -> 信阳东(Xinyang East) -> 武汉(Wuhan) -> 长沙南(Changsha South) -> 衡阳东(Hengyang East) -> 广州南(Guangzhou South).

​车站名称。列车停靠站点名称。运行路线顺序为：北京(Beijing) -> 北京西(Beijing West) -> 石家庄(Shijiazhuang) -> 郑州东(Zhengzhou East) -> 信阳东(Xinyang East) -> 武汉(Wuhan) -> 长沙南(Changsha South) -> 衡阳东(Hengyang East) -> 广州南(Guangzhou South)。

Format:String (e.g., 北京西, 广州南).

Importance:Defines the sequence and location of stops. 广州南is the final station and prediction target location.

​到达日期 (Arrival_Date): The date (year-month-day) the train was scheduled to arrive at this station.

​到达日期。列车计划到达该车站的日期（年-月-日）。

Format:Date (YYYY-MM-DD).

​到达时间 (Arrival_Time): The scheduled time of day (hour:minute) the train was planned to arrive at this station.

​到达时间。列车计划到达该车站的时刻（时:分）。

Format:Time (HH:MM, 24-hour).

​出发日期 (Departure_Date): The date (year-month-day) the train was scheduled to depart from this station. (Note: In the sample data, 到达日期 (Arrival_Date)and 出发日期 (Departure_Date)are always the same, as no journeys span midnight).

​出发日期。列车计划从该车站出发的日期（年-月-日）。(注意：在样本数据中，到达日期 (Arrival_Date)和 出发日期 (Departureure_Date)总是相同，因为无跨午夜旅程）。

Format:Date (YYYY-MM-DD).

​出发时间 (Departure_Time): The scheduled time of day (hour:minute) the train was planned to depart from this station.

​出发时间。列车计划从该车站出发的时刻（时:分）。

Format:Time (HH:MM, 24-hour).

​延误分钟 (Delay_Minutes): The ​target variable​ (present only in train.csv). The actual observed delay, in minutes, upon ​arrival​ at this station (e.g., 2= 2 minutes late). A value of 0indicates on-time arrival. ​For the test.csv, this value is MISSING for the final station (广州南/ Guangzhou South) and must be predicted by the model.​​ Note: The 出发时间 (Departure Time)may also be implicitly delayed if the train arrives late, but only the ​arrival delay at each station (especially the final station)​​ is explicitly recorded in this column and is the prediction target.

​延误分钟​：​目标变量​（仅在 train.csv中存在）。列车到达该车站时的实际观察到的延误分钟数（例如，2表示晚点 2 分钟）。0表示准点到达。​在 test.csv中，终点站 (广州南/ Guangzhou South) 的这一值是缺失的，必须由模型预测。​​ 注意：出发时间 (Departure Time)可能会因列车到达晚点而隐性延误，但此列明确记录的是各站（尤其是终点站）的到达延误，且是预测目标。

Format:Integer (can be 0or positive, negative values unlikely but theoretically possible if early arrival).

Importance:​This is the prediction target.​​

​distance: The ​cumulative distance traveled​ (in kilometers) from the origin station (北京/ Beijing)to the current station. This value increases sequentially along the route.

​距离​：从始发站 (北京/ Beijing) 到当前车站的累计行驶距离​（单位：公里）。该值沿路线依次增加。

Format:Integer (e.g., 0, 9, 290, 2253).

Importance:Key feature indicating progress along the route.