data-analysis

Hypothesis testing in Python requires validating normality and homoscedasticity before selecting parametric tests. This analysis skill automatically checks distribution assumptions and switches to non-parametric alternatives like Mann-Whitney or Kruskal-Wallis when assumptions fail, ensuring defensible statistical conclusions.

Calculating effect sizes with confidence intervals requires computing Cohen's d or eta-squared alongside p-values to interpret significance with magnitude. This skill quantifies evidence by reporting effect sizes, ensuring statistical significance is evaluated with practical importance for publication-ready documentation.

Generating publication-quality statistical plots from CSV files involves using pandas for data ingestion and seaborn/matplotlib for visualization. This skill produces journal-ready static figures with accessible styling and export-ready formats, plus optional interactive plotly plots for exploratory analysis.

Scipy supports reproducible research workflows by enabling assumption-checked ANOVA and t-tests. This skill uses scipy to validate test assumptions before choosing parametric tests, automatically switching to non-parametric alternatives when needed, and outputs reproducible documentation with every analysis.

Non-parametric tests like Kruskal-Wallis should be used instead of ANOVA when normality or homoscedasticity assumptions fail. This skill assesses distribution assumptions during exploratory analysis and automatically selects the appropriate non-parametric alternative to ensure valid statistical inference.

Exploratory data analysis and statistical inference on observational datasets can be performed using pandas and numpy. This skill handles data ingestion, cleaning, distribution assessment, and hypothesis testing, applying structured statistical analysis for inference and effect quantification across grouped comparisons.