Discord

Checklist USA-NA-AIO 2026 Round 2 · Task 5

Mixed Functions Parameter Regression

For groups of 400 noisy points drawn from four function families, estimate all 11 parameters of the functions.

  • Latent-structure regression

The task

Problem 5 (50 points, open-ended). Each group contains 400 (x, y) points, 100 from each of an exponential (a0·e^{a1 x} + a2), a logarithmic (b0 ln x + b1), a trigonometric (c0 sin(c1 x + c2) + c3) and a linear (d0 x + d1) function, with Gaussian noise; the function of each point is not given.

train.csv (400 groups) and train_params.csv give points and true parameters; test.csv has 200 groups. The baseline clusters each group with k-means (k = 4) and fits the four function types to the clusters.

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, train_params.csv, test.csv (columns group_id, x, y) and a sample submission.
You submit
A CSV with the 11 parameters per test group, a notebook and a .docx report, named mix_functions_LastName_FirstName_SchoolName.
Scoring
RMSE over all predicted parameters of all groups; final score = (X_baseline − X)/(X_baseline − X_best) × 100%, 0 if worse than the baseline.
Format
2026 USA-NA-AIO Round 2, Day 2 (5 April 2026).

Details

Year
2026, Proctored
Round
Round 2 · Task 5
Language
English
License
Not stated by the source