Ch.02 Mini Project 05 — Categorical Relationship Analyzer¶

Concept: Bivariate analysis involving qualitative attributes — contingency tables, mosaic plots, grouped box plots, and jitter scatter plots for ordinal pairs.

What it does¶

categorical_analyzer.py covers all three categorical bivariate scenarios from the lecture:

Attribute pair	Analysis	Visualization
Qual + Qual	Contingency table	Mosaic plot
Qual + Quant	Grouped summary stats per category	Grouped box plots
Ordinal + Ordinal	Spearman's rho	Jitter scatter plot

Dependencies¶

pip install matplotlib

Usage¶

# Built-in demo (Ch.2 Friends dataset — Company, Gender, Weight, Height)
python categorical_analyzer.py

# Analyze your own CSV (auto-detects column types)
python categorical_analyzer.py my_data.csv

What each analysis shows¶

1. Contingency Table (Qual + Qual)¶

A matrix of joint frequencies with row totals, column totals, and grand total. Both absolute counts and relative percentages are shown.

  Contingency Table: Company (rows) vs. Gender (cols)
  =====================================================
  Company          F         M     Total
  -------------------------------------------------
  Bad       4 (28.6%)  4 (28.6%)        8
  Good      2 (14.3%)  4 (28.6%)        6
  -------------------------------------------------
  Total             6         8       14

2. Mosaic Plot (Qual + Qual)¶

A rectangle for each cell of the contingency table. Width = column marginal proportion, height = conditional proportion within the column. Areas are directly proportional to relative joint frequency.

3. Grouped Box Plots (Qual + Quant)¶

One box plot per category, all on the same axis. Reveals whether the distribution of the numeric attribute differs meaningfully between groups.

4. Jitter Scatter Plot (Ordinal + Ordinal)¶

Adds a small random offset to each point. Without jitter, all identical (rank, rank) pairs overlap as a single dot — making the cloud invisible. Side-by-side comparison of with/without jitter shows why the technique matters.

Key Concepts from Ch.2 Applied¶

Concept	Where it appears
Contingency table	`build_contingency_table()`, `print_contingency_table()`
Mosaic plot (area = relative freq)	`plot_mosaic()`
Grouped box plots	`plot_grouped_boxplots()`
Jitter effect	`plot_jitter_scatter()`
Spearman rho (ordinal pair)	`spearman_rho()` — same as project-04
Five-number summary	`five_num_summary()`

Limitations & Future Ideas¶

Extension: add chi-squared (χ²) test of independence to the contingency table output.
Extension: add a Cramér's V statistic (strength of association for nominal-nominal pairs).
Extension: support ordered categories (e.g., Low < Medium < High) in mosaic and contingency tables.