Prediction Model of Motivators and Demotivators of Integrating Large Language Models in Software Engineering Education: An Empirical Study

This replication package accompanies the study “Prediction Model of Motivators and Demotivators of Integrating Large Language Models in Software Engineering Education: An Empirical Study” and provides all materials necessary to reproduce the empirical analyses and optimization results reported in the paper.

Overview

The package enables full reproducibility of:

Data cleaning and preprocessing

Descriptive statistical analysis of all factors

Binary outcome construction (“Success” based on LLM familiarity thresholding)

Baseline predictive modeling using Naive Bayes and Logistic Regression

Genetic Algorithm (GA) optimization for:

Global/locao allocation cost/effort) across all the factors

Theme-wise allocation across 8 thematic groups

All analyses were implemented in Python 

Contents of the Package

1. Dataset Files

Responses.csv – Original raw survey dataset (141 responses)

Responses_Cleaned_Replication.csv – Cleaned dataset used for replication

Responses_126_Complete_for_Analysis.csv – Filtered dataset (126 complete responses used in modeling)

2. Analysis Code

Fully executable Python script/notebook reproducing:

Data preprocessing

Likert-scale normalization

Construction of the binary outcome variable

Model training and evaluation

Global and theme-wise GA optimization

3. Output Files

best_solution_table_global_GA.csv/.xlsx – Global optimal allocation (all factors)

theme_results_summary.csv/.xlsx – Summary of theme-wise optimization results

theme_best_allocations_*.csv/.xlsx – Optimal allocations per theme

Combined tables for reporting and LaTeX export