additional_features
["Missingness workflow with a 5% threshold plus Little’s MCAR test decision path", "Multicollinearity rules: recommend removal at VIF>5 and require removal at VIF>10", "PLS workflow: mandatory standardization, CV-based component selection, and reporting VIP/Q²/RMSECV", "A 5-step interpretation template (purpose→methods→results→implications→comparison to prior work)"]
example_commands
["Analyze the ANPOR dataset: fit a 3-step hierarchical OLS model predicting sustainable actions, and report missing/outliers/VIF/residual diagnostics.", "My Poisson model has dispersion = 2.0. Determine overdispersion, refit using negative binomial, and write an interpretation paragraph.", "The dependent variable has skewness 2.8. Should I log-transform it? If yes, show how to back-translate effects to the original scale.", "Run PLS with k-fold CV to choose components, then summarize results foc
gpt_id
g-68e00f18911081919ca2a1fec0491068
ideal_use_cases
["Document a standardized preprocessing/diagnostics workflow for survey or observational regression analyses", "Run hierarchical regression by adding variable blocks and comparing ΔR² across steps", "Design moderated regression with interaction terms (e.g., platform/group moderators)", "Assess overdispersion in Poisson models and propose NB/ZIP/ZINB alternatives", "Conduct PLS regression with cross-validated component selection and VIP-focused interpretation"]
limitations
["Recommendations are rule-based; final decisions must reflect the study context (measurement, sampling, causal assumptions)", "If raw data quality is poor (coding errors, sampling bias, low response reliability), interpretability and trustworthiness are limited", "Auto-formatting to specific journal/institution reporting templates may require additional templates"]
target_users
["Social science & policy researchers", "Graduate students writing theses/papers", "Data analysts doing regression-heavy work"]