Python and ChatGPT: AI-Powered Data Visualization for Teachers

This study explores the synergy of Python and ChatGPT's intuitive accessibility to empower non-programming educators in making data-driven decisions. Using a Quantitative Ethnography (QE) approach, we analyzed data discourse of behavior of twenty teachers from three private schools through AI-guided prompts. Qualitative coding first identified teachers' difficulties and strategies. Epistemic Network Analysis (ENA) was then employed to generate network models, revealing how these codes co-occur. We observed a shift in teachers' frameworks over a three-week period. In Session 1, the conceptual focus was on technical hurdles, with strong connections linking "Prompt engineering," "Data cleaning & preparation," and "Syntax errors". This indicates a discourse dominated by a linear, technical problem-solving approach. The group centroid for this session was located on the left side of the conceptual space. By Session 2, a shift occurred with a large effect size (Cohen's d=0.93). The restructuring shows that as teachers overcame initial technical challenges, their acumen elevated from low-level tool mechanics to higher-order pedagogical application. The findings provide an account of a transition from a conceptual space dominated by technical difficulties to one focused on the strategic, critical, and pedagogical application of data vis-à-vis their professional practice. While the study is limited by the exclusion of data storytelling context, it highlights the need for additional design features to help teachers engage meaningfully with data.