Checklist AICC 2025 Round 0 · Task 1
Deceptive Points
Fit a regression model on data that mixes genuine entries with deliberately corrupted ones, so that it predicts the true linear trend on a clean test set.
The task
A teacher built a dataset in which exam scores generally increase linearly with study effort. Some students added conflicting entries to make it appear that more effort does not improve scores.
The training data contain both the teacher's original rows and the students' corrupted rows, with four numerical features and a numerical target. The test set contains only valid teacher entries.
Contestants predict the true expected score for each test row.
Abridged by SOTA from the official materials. The official statement has the exact rules, and it wins wherever this summary differs.
At a glance
- You get
train.csv(feature1–feature4, target) andtest.csv(ID, feature1–feature4).- You submit
submission.csvwith columns ID and Target.- Scoring
- Mean squared error against the teacher-only targets; lower is better.
- Rules
- Any model may be used (for example linear regression, tree ensembles or neural networks), with any preprocessing of the four features.
- Individual participation (maximum team size 1); at most 15 submissions per day.
- AICC contest rules (stated on each Kaggle rules page, not enforceable): no use of LLMs for writing code or getting task ideas; no internet use other than official library documentation and the contest platform; no communication with anyone during the contest; clarifications only via the #clarification-requests channel on the AICC Discord server.
- Format
- AICC Round 0, online on Kaggle, 4 Oct 2025 18:00 UTC – 5 Oct 2025 18:00 UTC.