Integrating Chemical Kinetics Inquiry with the Data Science Front End: Primary Data Collection, Preprocessing, and Exploratory Data Analysis

Abstract This study proposes an integrated framework that connects chemical kinetics inquiry with the front end of data science, comprising primary data collection, data preprocessing, and exploratory data analysis (EDA). While many data science educational programs rely on repository-based, ready-to-use data, this study emphasizes the educational value of acquisition-required data generated through students’ own experiments. Eighty-seven high school students participated in a four-week inquiry on the reaction between sodium bicarbonate and citric acid. The activity consisted of three inquiry stages: collecting primary mass-change data under varied NaHCO3 mass fraction and temperature conditions, preprocessing the data by cleaning records and transforming mass loss into a reaction-rate indicator (ΔwCO2/Δt), and conducting EDA using Orange 3. Students conducted three exploratory analyses: visualizing ΔwCO2/Δt over time, examining initial reaction rate versus NaHCO3 mass fraction under controlled temperature conditions, and examining initial reaction rate versus temperature at a fixed NaHCO3 mass fraction. This process helped students connect data patterns in the reaction-rate experiment with kinetic ideas such as reactant depletion, concentration dependence, and temperature dependence expressed through Arrhenius-type behavior. Quantitative analysis using paired-sample t tests showed significant pre-to-post increases in students’ overall interest in science (p = 0.010) and overall attitudes toward science (p = 0.003), with small effect sizes (dz = 0.28 and 0.33). Subdomain-level changes, including a decrease in science-learning anxiety, were treated as exploratory. These findings suggest that foregrounding data generation, preprocessing, and EDA within authentic chemical inquiry can provide a meaningful foundation for students’ data literacy in science and can later be extended toward predictive modeling and machine-learning activities.