Sport-agnostic skills - Python, pandas, plotting and data cleaning - that every other tutorial builds on.
No tutorials match that filter.
Install Python, a code editor, and the core data libraries the right way, then prove your setup works by running a tiny script that imports pandas and matplotlib.
Load, filter, sort, group, and join a real standings table. The twelve pandas operations here cover the vast majority of everyday sports-data work.
Go from a DataFrame to a clean, labeled bar chart with matplotlib. Learn the figure/axes model, titles, and how to save a chart you'd actually publish.
Endpoints, parameters, JSON, status codes - learn to read API docs by pulling live standings from the public NHL API and shaping the response into a DataFrame.
Before you chart anything, meet your data. Use describe() to summarize a column in a single line, then draw a histogram to see the distribution those numbers only hint at.
Real data rarely arrives in one table. Learn pandas merge/join: combine a record table and a scoring table on a shared key, choose the right join type, and chart a column that lived in neither alone.
Season totals hide the rates that matter. Build new columns - runs per game, run differential per game, win percentage - from existing ones with vectorized math, then chart them.
Go beyond arithmetic on columns: use .apply() to run any function per row and .map() to translate codes to labels, turning raw numbers into the categories your analysis needs.
pd.cut() slices a continuous column into fixed, equal-width ranges you define, turning a raw number into labeled tiers. It's the companion to qcut() (equal counts) - here you control the edges and the bands can come out uneven.
Mixed types, stray whitespace, duplicate rows, percent signs in numbers - fix the data problems that derail real sports projects, with before-and-after proof.
Go beyond groupby: build pivot tables, reshape between wide and long with melt and stack, and turn raw rows into the summary tables analysts actually present.
Pull a team's game log, sort by date, and use rolling windows and cumulative sums to measure form and momentum - the foundation of every 'hot streak' chart.
Take a default matplotlib chart and make it presentable: custom colors, direct labels, annotations that call out the story, cleaned-up axes, and a high-resolution export.
Learn matplotlib's subplots to build small multiples - a grid of tiny charts that share axes so you can compare many teams or groups at a single glance.
Runs scored, runs allowed, and run differential live on different scales. Convert each to a z-score - standard deviations from the mean - so a single number means the same thing in every column, then read it off a diverging bar chart.
Where a histogram bins and smooths, the ECDF is exact. Sort the values, assign each its running share, and plot a staircase you can read 'what fraction of teams fall below X?' straight off - the most underrated distribution chart there is.
Runs scored and win percentage live on wildly different scales. Use ax.twinx() to give a chart a second y-axis so both series fit on one plot - and learn when a dual axis clarifies and when it quietly misleads.
A capstone: ask one question - how big is home advantage? - across five leagues, reusing everything you've learned, and put the answers on a single comparison chart.
Quantify a relationship instead of eyeballing it: compute a correlation, fit a regression line with numpy, and report R-squared - using NHL goal differential to predict points.
Reconstruct the standings week by week from a season of game results and draw a bump chart - the elegant ranked-line visual that shows every rise and fall.