diff --git a/playground/example_ab_test.html b/playground/example_ab_test.html
new file mode 100644
index 00000000..be1ec074
--- /dev/null
+++ b/playground/example_ab_test.html
@@ -0,0 +1,260 @@
+
+
+
+
+
+ tsb — A/B Test — Examples
+
+
+
+
+
+
Initializing playground…
+
🧪 A/B Test Results
+ Scenario: A product manager just shipped a new checkout button (variant B) to half of users. Compare conversion rates and order values between the control (A) and the variant (B).
+ Skills: groupby().agg(), describe, lift calculation, boolean masks.
+
+
+
1 · Conversion rate by variant
+
Each row is one user session: which arm they were in, whether they converted, and order value.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Order value distribution per variant
+
Use describe to compare full distributions, not just means.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
👥 Customer Signup Cohorts
+ Scenario: A SaaS growth team wants to know how many customers signed up each month, the cumulative customer base, and which cohort grew fastest.
+ Skills: groupby, cumsum, pctChangeSeries.
+
+
+
1 · Monthly signups & cumulative growth
+
Group raw signup events by their cohort month, then take a running total.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Visualise cohort growth
+
Bar chart of new customers per month.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
🪵 Server Access Log Analysis
+ Scenario: An on-call engineer wants to know how many 5xx errors hit the API per hour during the last incident, broken down by status code class.
+ Skills: column transforms via Series.map, groupby, pivot, error rates.
+
+
+
1 · Parse and bucket access logs
+
Each row is one HTTP request. Extract the hour and the status class (2xx/4xx/5xx).
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Error rate per hour & alert threshold
+
Compute the 5xx error percentage per hour and flag any hour above 30%.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
🏷️ Product Pricing Tiers
+ Scenario: An e-commerce merchandiser wants to bucket the catalogue into 4 price tiers (Budget / Mid / Premium / Luxury) and see the count and average margin per tier.
+ Skills: cut with custom labels, valueCounts, groupby.
+
+
+
1 · Bucket products into price tiers
+
Use fixed bin edges with cut and named labels.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Avg margin by tier
+
Strategy question: do premium products carry meaningfully higher margins?
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
💰 Sales Dashboard
+ Scenario: You're an analyst at a regional retail chain. Q1 sales just landed in a CSV. Find top-performing regions and products, then visualize revenue with a quick ASCII bar chart.
+ Skills you'll use: readCsv, groupby().agg(), sortValues, nlargestDataFrame.
+
+
+
1 · Load the sales CSV
+
A typical first step: parse a CSV string into a DataFrame and inspect the schema.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Compute revenue & rank regions
+
Add a derived revenue column, then group by region.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
3 · Top products & best single order
+
Use groupby + nlargestDataFrame to surface the headline numbers.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
⚽ Sports League Standings
+ Scenario: An amateur football league played a round-robin. Build the league table from match results: wins, goal difference, points, and final rank.
+ Skills: concat, groupby().agg, rankSeries, sortValues.
+
+
+
1 · Build the standings table
+
Each match contributes two rows to the per-team view (one for each team).
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
📈 Stock Returns Analysis
+ Scenario: A junior quant wants to inspect a price history: compute daily returns, a 5-day rolling mean and volatility, and detect a simple moving-average crossover signal.
+ Skills: pctChangeSeries, Series.rolling().mean/std, derived columns, basic signal generation.
+
+
+
1 · Daily prices and returns
+
Compute daily percentage returns from raw closing prices.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Rolling stats & SMA crossover signal
+
A common workflow: fast SMA (3-day) crossing a slow SMA (5-day) is a buy signal; the reverse is a sell signal.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
📊 Survey Cross-tab Analysis
+ Scenario: You ran a 30-person survey asking which programming language people prefer, broken down by experience level. Build a contingency table and a percentage breakdown.
+ Skills: crosstab, normalisation modes ("index", "columns", "all").
+
+
+
1 · Build the crosstab
+
Each row in responses is one survey reply.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Row percentages
+
Row-normalise to see what each cohort actually prefers.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
🌦️ Monthly Weather Trends
+ Scenario: A climate journalist has a year of daily temperature observations and wants monthly averages, the hottest month, and a quick visual of the warming curve.
+ Skills: datetime parsing via the dt accessor, groupby, idxmaxSeries.
+
+
+
1 · Aggregate daily readings into monthly means
+
Synthetic but realistic year of daily highs in °C.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Visualise the seasonal curve
+
An ASCII bar chart of monthly averages — a quick health-check before reaching for proper plotting.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
+
Initializing playground…
+
🌐 Web Analytics Pivot Table
+ Scenario: A digital marketer has a stream of pageview events tagged with traffic source and device type. Pivot the data to see how each source performs across devices.
+ Skills: pivotTable, multiple aggregations, totals.
+
+
+
1 · Pivot pageviews by source × device
+
Sum pageviews into a 2-D table.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
+
2 · Add row totals + bar chart of sources
+
Marketing wants the bottom line: total pageviews per source.
+
+
+
+
+
Click ▶ Run to execute
+
Ctrl+Enter to run · Tab to indent
+
+
📚 Real-World Examples
+ Each example is a complete, real-world workflow built with tsb — the kind of analysis you'd typically reach for pandas to do. Click any card to open its interactive page; every code block runs live in your browser.
+ The dataset in each example is small, inline, and editable — change the numbers and re-run to see the analysis update instantly.
+
+
+
+ 💰
+ Sales Dashboard
+ Scenario: You're an analyst at a regional retail chain. Q1 sales just landed in a CSV. Find top-performing regions and products, then visualize revenue with a quick ASCII bar chart.
+
+
+
+ 📈
+ Stock Returns Analysis
+ Scenario: A junior quant wants to inspect a price history: compute daily returns, a 5-day rolling mean and volatility, and detect a simple moving-average crossover signal.
+
+
+
+ 🌦️
+ Monthly Weather Trends
+ Scenario: A climate journalist has a year of daily temperature observations and wants monthly averages, the hottest month, and a quick visual of the warming curve.
+
+
+
+ 👥
+ Customer Signup Cohorts
+ Scenario: A SaaS growth team wants to know how many customers signed up each month, the cumulative customer base, and which cohort grew fastest.
+
+
+
+ 📊
+ Survey Cross-tab Analysis
+ Scenario: You ran a 30-person survey asking which programming language people prefer, broken down by experience level. Build a contingency table and a percentage breakdown.
+
+
+
+ 🪵
+ Server Access Log Analysis
+ Scenario: An on-call engineer wants to know how many 5xx errors hit the API per hour during the last incident, broken down by status code class.
+
+
+
+ 🧪
+ A/B Test Results
+ Scenario: A product manager just shipped a new checkout button (variant B) to half of users. Compare conversion rates and order values between the control (A) and the variant (B).
+
+
+
+ 🌐
+ Web Analytics Pivot Table
+ Scenario: A digital marketer has a stream of pageview events tagged with traffic source and device type. Pivot the data to see how each source performs across devices.
+
+
+
+ ⚽
+ Sports League Standings
+ Scenario: An amateur football league played a round-robin. Build the league table from match results: wins, goal difference, points, and final rank.
+
+
+
+ 🏷️
+ Product Pricing Tiers
+ Scenario: An e-commerce merchandiser wants to bucket the catalogue into 4 price tiers (Budget / Mid / Premium / Luxury) and see the count and average margin per tier.
+
+
pandas for TypeScript
+ 📚 Real-World Examples
+
+
+
+
Ten complete, real-world workflows running interactively in the browser — sales dashboards, stock returns, A/B tests, web analytics, league tables, and more.
+
✅ New — start here
+
+