Update data-analysis materials for "Python for Data Analysis" refresh

[realpython-bot]

Syncs the data-analysis/ materials with the refreshed Python for Data Analysis tutorial and its updated dependency stack.

Dependencies

Added data-analysis/requirements.txt pinning the versions from the tutorial update:

pandas==3.0.3
matplotlib==3.10.9
scikit-learn==1.9.0
openpyxl==3.1.5
pyarrow==24.0.0
lxml==6.1.1

(The tutorial targets Python 3.14.)

Code changes (both notebooks)

• Currency cleanup regex now uses a raw string and also strips whitespace: .replace("[$,]", "", regex=True) → .replace(r"[$,\s]", "", regex=True). The source CSV stores figures with surrounding spaces (" $1,000.00 "), so this makes the cleanup explicit rather than relying on astype() to trim.
• film_length suffix removal now removes the leading space too: .str.removesuffix("mins") → .str.removesuffix(" mins").
• read_html() now passes a browser User-Agent via storage_options={"User-Agent": "Mozilla/5.0"}, since Wikipedia now returns HTTP 403 Forbidden without one (findings notebook).

Regenerated artifact

• james_bond_data_cleansed.csv was regenerated under pandas 3.0. In pandas 3.0, .combine_first() no longer alphabetically sorts the result columns, so the cleansed file now preserves the logical source column order. The data values are unchanged — only the column order differs.

Verification

Ran the full pipeline end-to-end on the pinned versions (Python 3.14, pandas 3.0.3, scikit-learn 1.9.0, matplotlib 3.10.9):

• Cleansing reproduces the dataset (25 rows, 0 nulls; typos and outliers fixed).
• Regression analysis still yields R-squared 0.79, best fit y = 1.6637x - 4.9276.
• film_length stats: min 106, max 163, mean 128.28, std 12.94 — matching the tutorial output.

🤖 Generated with Claude Code

[martin-martin]

Closed in favor of #780