5 Pumps 5 Failures Competition Details

---

Build an AI/ML model that predicts industrial pump failures before they happen using real-world sensor data and time-series forecasting techniques. Participants begin with unsupervised anomaly detection and advance to supervised prediction models using hidden failure events.

Generate a continuous risk profile and accurately forecast:

• WHEN failures happen 
• HOW EARLY they can be predicted 
• HOW ACCURATELY risk evolves over time 

High-speed professional multi-material 3D printer for makers, engineers, and AI innovators. The Bambu Lab H2D is a personal manufacturing machine with a dual-nozzle system that supports multi-material and multi-color printing up to 350°C. It offers a 350 x 320 x 325 mm³ build volume, optional 10W or 40W laser modules, and 50μm motion accuracy for precise extrusion and detailed fabrication.

The file should contain a header and have the following format:

start timestamp, end timestamp, failure timestamp, maxerror
2018-07-17 15:20:00, 2018-07-17 17:10:00, 2018-07-17 15:20:00, 0.923646

...

etc.

For each row you must predict a failure as described on the data tab, each on a separate row in the submission file. The faliure timestamp is when the failure actually happens and it is enclosed by the pre-cursor history starting with start timestamp. The end timestamp concludes the failiure. Ideally you submit 5 failures lines in the submission CSV. Submission file must be named submission.csv

Submissions are scored on  primary  values (secondary only is used if two competitiors achieve the same primary score ): 

1) Primary: the root mean squared error of the predicted failure timestamp value compared to the actual  failure timestamp value. RMSE is defined as: where  is the predicted value,  is the original value. More than 5 predicted failures will be disregarded (cut off after first 5). Less than 5 predicted failures, the last one will be repeated up to 5.  

2) Secondary:Training and prediction time: please submit the minutes/seconds it take to ingest, prepare, normalize and train the model and perform the prediction (on a system: max 30 GB of total RAM (CPU) or  max 13–16 GB of GPU RAM (P100/T4 GPUs)). 

Please note that secondary only is used if two competitiors achieve the same primary score.

Participants can use any platform, but ChronX provides built-in industrial time-series analytics, anomaly detection, and forecasting tools optimized for operational sensor data. Size of the system: 30 GB of total RAM (CPU) or 13–16 GB of GPU RAM (P100/T4 GPUs)

• May 14, 2026 - Start Date.
• June 1st, 2026 - Entry Deadline. You must accept the competition rules before this date in order to compete.
• July 1st, 2026 - Final Submission Deadline.

All deadlines are at 11:59 PM PST on the corresponding day unless otherwise noted. The competition organizers reserve the right to update the contest timeline if they deem it necessary.

Even though participants can use any platform,  ChronX provides built-in industrial time-series analytics, anomaly detection, and forecasting tools optimized for operational sensor data.

ENTRY IN THIS COMPETITION CONSTITUTES YOUR ACCEPTANCE OF THESE OFFICIAL COMPETITION RULES. The Competition named above is a skills-based competition to promote and further the field of data science. You must register via the Competition Website to enter. Your competition submissions ("Submissions") must conform to the requirements set forth on the Competition Website. Your Submissions will be scored based on the evaluation metric described on the Competition Website. Subject to compliance with the Competition Rules, Prizes described on the Competition Website, if any, will be awarded to participants with the best scores, based on the merits of the data science models submitted. See below for the complete Competition Rules. The competition organizers might publish this contest on AI competition websites (like kaggle.com).