Tutorial: Lessons Learned for Behavior Analysts from Data Scientists

Neely et al. (2024) wrote a how-to guide for behavior analysts who want to use data-science tools.

The authors explain how to clean, protect, and format large sets of session data.

They show where to find a data-science partner and which steps turn raw numbers into machine-learning models.

The paper does not test a treatment. Instead, it lists lessons the authors learned while working with big data sets.

Key advice: tidy data first, lock up private information, and let the algorithm look for hidden patterns.

Slanzi et al. (2024) also published a 2024 tutorial, but on the Countee app for single-case data entry. Both papers push digital tools, yet Neely aims at huge agency-wide files while Slanzi targets one-client sheets.

Moeyaert et al. (2020) teach multilevel meta-analysis of single-case designs. Neely extends that idea by showing how machine-learning models can spot trends across thousands of cases without hand-coding each graph.

Burney et al. (2023) argue for adding qualitative interviews to behavior analysis. That sounds opposite to Neely’s number-heavy path, but the two calls fit together: numbers tell you what happens, interviews tell you why clients care.

If you store every trial in one messy spreadsheet, you already have big data. This paper gives you a checklist to make that spreadsheet model-ready. Start small: pick one month of session data, remove extra spaces, add a client-ID column, and save a locked copy. Then ask a local data-science student to run a simple clustering script. You may discover response patterns you never saw by eye.