Skip to content

UC Berkeley course transaction dataset and publicly released analysis code

Product or Programme · All products and programmes
Evidence · no claims cited

Description

A UC Berkeley dataset of over 11 million timestamped student course-enrollment transactions from Fall 2016 through Spring 2022, released with public analysis code by Borchers et al.

Components

  • Multi-attempt student code submission dataset from a Berkeley introductory programming course with AI tutor feedback: A dataset of student code submission streams from an introductory programming course at UC Berkeley (~900 students per semester), spanning Spring 2024 and Spring 2025 and covering the first five Python homework assignments. Each attempt records the student's code, autograder results, and natural-language feedback from an LLM-based tutor for consenting students. The article reports partitioning "using Spring 2025 for training (445 students, 2,022 submission streams, 6,911 total submissions) and Spring 2024 for testing (479 students, 1,546 submission streams, 6,316 total submissions)," with test_OP (5,262 submissions) and test_NP (1,054 submissions) generalization subsets. The data received IRB approval and identifiers were anonymized. (Rose Niousha et al. (2026))

Claims

Key Sources

  • Borchers, C., Xu, Y., & Pardos, Z. A. (2025). Workload Overload? Late Enrollment Leads to Course Dropout. Journal of Educational Data Mining, Volume 17, No 1. https://github.com/CAHLR/enrollment-procrastination-edm
Page metadata
Field Value
type product
id uc-berkeley-course-transaction-dataset-and-publicly-released-analysis-code
title UC Berkeley course transaction dataset and publicly released analysis code
description A UC Berkeley dataset of over 11 million timestamped student course-enrollment transactions from Fall 2016 through Spring 2022, released with public analysis code by Borchers et al.
status review
moved_from element/uc-berkeley-course-transaction-dataset
product_kind dataset
generated by process:sweep-kinds on 2026-10-10
trust unverified — no one has confirmed this page's content
sources 1 — borchers-2025

Frontmatter is the page's machine-readable record — see the Schema & Guide.