Petronel the cyclist

Petronel the Cyclist

Problem Description

In order to maintain his health, Petronel has decided to be more athletic. To this end, he has started cycling regularly, both to work and for recreational purposes. He uses a dedicated application to monitor the rides he takes and is interested in obtaining statistics about his rides and observing how he manages to combine the useful with the pleasant, categorizing rides according to their purpose.

After obtaining a dataset that describes all the rides he has taken since he started cycling, Petronel wants to find out what is the average speed at which he travels, according to the calendar month. Additionally, Petronel has started labeling rides, dividing them into: rides where he traveled between home and work, without making any detour; rides where he traveled between home and work, also making a detour for recreational purposes; and purely recreational rides. Because labeling is cumbersome, he needs a program that automatically performs the labeling of the remaining rides (and future ones) with reasonable accuracy.

Training Dataset

The file dataset_train.csv presents the following columns in the header:

• Activity ID: unique ID of the ride
• Activity Date: date and time of the ride's start in GMT timezone, in the format Aug 23, 2025, 4:55:36 PM
• Distance: distance traveled, expressed in kilometers,
• Elapsed Time: total duration of the ride, including rest time, expressed in seconds
• Moving Time: actual duration of cycling, expressed in seconds
• Starting Latitude, Starting Longitude: geographical coordinates of the starting point
• Finish Latitude, Finish Longitude: geographical coordinates of the stopping point
• Label: label applied to the ride, with the following meanings:
  • COMMUTE: commute ride without detour;
  • LEISURELY_COMMUTE: commute ride with detour;
  • PURE_LEISURE: purely recreational ride.

Evaluation Dataset

The file dataset_eval.csv presents the same fields as the training data file, except for the Label field, which needs to be determined.

Requirements

Requirement 1

Determine the average travel speed for each month of the year, expressed in km/h, according to the data available in the training dataset.

Requirement 2

Perform labeling of the rides present in the file dataset_eval.csv, starting from the labels of the records in the file dataset_train.csv.

Evaluation

For requirement 1: 2.5 points for each correctly calculated average speed for each of the 12 months of the calendar year - in total, 30 points.

For requirement 2: 0.82 points for each 1% of the dataset labeled correctly, up to a maximum threshold of 70 points. In other words, ~15% incorrect labeling is tolerated.

Submission File

The output file is in CSV format, with the following header:

1subtaskID,Answer1,Answer2

For requirement 1, exactly 12 rows, one for each month of the year, as follows:

11,<month>,<average speed>

The month will be expressed through the three-letter abbreviation (Jan, Feb etc.) and the average speed will be expressed as a real number with exactly 5 decimal places after the comma, rounded down.

For requirement 2, one row for each record present in the evaluation file, as follows:

12,<ride id>,<label>

The ride ID is taken directly from the evaluation file, and the label has one of the values described above.

Starting kit

Find at this link a zip archive containing the training and evaluation files, as well as an example of a valid submission file, which scores 1 point.