# Nushell Carts Combo

> Source: <https://gist.github.com/paul-d-ray/c8f85c7f55afa987a34247f6354574d1>
> Published: 2026-07-15 17:28:09+00:00

| #!/usr/bin/env nu | |
| # PR 2026-07-15 11:23:32 | |
| # https://mavenanalytics.io/data-drills/cart-combos | |
| #Your dataset contains one year of point-of-sale transactions | |
| # from a grocery store, including approximately 88,000 line items | |
| # across 20,000 customer orders. | |
| # | |
| #Your task is to identify the five product pairs that are purchased | |
| # together most often. | |
| # | |
| #Two products are "purchased together" if they appear in the same | |
| # transaction | |
| # | |
| #A pair must consist of two different products | |
| # | |
| #Product pairs are unordered, meaning (A, B) and (B, A) should | |
| # be treated as the same pair | |
| # | |
| #Count each pair once per transaction, regardless of the quantity | |
| # purchased | |
| # ********************************************************************** | |
| # code from gemini (36 seconds) | |
| # Load the transactions and select only the relevant columns | |
| open grocery_transactions.csv | |
| | select transaction_id product_name | |
| # Remove duplicate products within the same transaction to count them once per transaction | |
| | uniq | |
| # Group the dataset by transaction ID | |
| | group-by transaction_id | |
| | values | |
| # For each transaction, generate all unique product pairs | |
| | each { |tx| | |
| let products = ($tx.product_name | sort) | |
| $products | enumerate | each { |item| | |
| let idx = $item.index | |
| let p1 = $item.item | |
| $products | skip ($idx + 1) | each { |p2| | |
| { product_a: $p1, product_b: $p2 } | |
| } | |
| } | flatten | |
| } | |
| | flatten | |
| # Count the occurrences of each unique pair | |
| | uniq --count | |
| # Sort by count descending and take the top 5 | |
| | sort-by --reverse count | |
| | first 5 | |
| ╭───┬────────────────────────────────┬───────╮ | |
| │ # │ value │ count │ | |
| ├───┼────────────────────────────────┼───────┤ | |
| │ 0 │ ╭───────────┬────────────────╮ │ 5358 │ | |
| │ │ │ product_a │ Salsa │ │ │ | |
| │ │ │ product_b │ Tortilla Chips │ │ │ | |
| │ │ ╰───────────┴────────────────╯ │ │ | |
| │ 1 │ ╭───────────┬─────────╮ │ 4483 │ | |
| │ │ │ product_a │ Bananas │ │ │ | |
| │ │ │ product_b │ Milk │ │ │ | |
| │ │ ╰───────────┴─────────╯ │ │ | |
| │ 2 │ ╭───────────┬─────────╮ │ 2970 │ | |
| │ │ │ product_a │ Bananas │ │ │ | |
| │ │ │ product_b │ Bread │ │ │ | |
| │ │ ╰───────────┴─────────╯ │ │ | |
| │ 3 │ ╭───────────┬─────────╮ │ 2874 │ | |
| │ │ │ product_a │ Bananas │ │ │ | |
| │ │ │ product_b │ Salsa │ │ │ | |
| │ │ ╰───────────┴─────────╯ │ │ | |
| │ 4 │ ╭───────────┬────────────────╮ │ 2857 │ | |
| │ │ │ product_a │ Bananas │ │ │ | |
| │ │ │ product_b │ Tortilla Chips │ │ │ | |
| │ │ ╰───────────┴────────────────╯ │ │ | |
| ╰───┴────────────────────────────────┴───────╯ | |
| #********************************************************************** | |
| # code from lumo.ai (30 seconds) | |
| # Top 5 product pairs purchased together most often | |
| open grocery_transactions.csv | |
| | select transaction_id product_name | |
| | group-by transaction_id | |
| | items { |tid, rows| | |
| let products = ($rows.product_name | uniq | sort) | |
| let n = ($products | length) | |
| if $n < 2 { [] } else { | |
| seq 0 ($n - 2) | |
| | each { |i| | |
| seq ($i + 1) ($n - 1) | |
| | each { |j| | |
| $"($products | get $i) × ($products | get $j)" | |
| } | |
| } | |
| | flatten | |
| } | |
| } | |
| | flatten | |
| | uniq --count | |
| | sort-by count --reverse | |
| | first 5 | |
| ╭───┬──────────────────────────┬───────╮ | |
| │ # │ value │ count │ | |
| ├───┼──────────────────────────┼───────┤ | |
| │ 0 │ Salsa × Tortilla Chips │ 5358 │ | |
| │ 1 │ Bananas × Milk │ 4483 │ | |
| │ 2 │ Bananas × Bread │ 2970 │ | |
| │ 3 │ Bananas × Salsa │ 2874 │ | |
| │ 4 │ Bananas × Tortilla Chips │ 2857 │ | |
| ╰───┴──────────────────────────┴───────╯ | |
| #********************************************************************** | |
| # nushell with polars (0.5 seconds) | |
| # from gemini | |
| # 1. Load the CSV file as a Polars LazyFrame and select relevant columns | |
| let df = (polars open grocery_transactions.csv | |
| | polars select [transaction_id product_name] | |
| # Remove duplicate products within the same transaction | |
| | polars unique --subset [transaction_id product_name] | |
| ) | |
| # 2. Prepare a clone of the DataFrame for the self-join | |
| let df_right = ($df | |
| | polars rename product_name product_b | |
| ) | |
| # Rename the original DataFrame's product column for matching | |
| let df_left = ($df | |
| | polars rename product_name product_a | |
| ) | |
| # 3. Perform the self-join, filter, group, count, sort, and fetch top 5 | |
| $df_left | |
| | polars join $df_right [transaction_id] [transaction_id] | |
| # Keep only pairs where product_a comes alphabetically before product_b | |
| | polars filter-with ((polars col product_a) < (polars col product_b)) | |
| # Group by the distinct product pairs | |
| | polars group-by [product_a product_b] | |
| # Count occurrences of each pair | |
| | polars agg [ | |
| ((polars col transaction_id | polars count) | polars as count) | |
| ] | |
| # Sort by count descending (Passing '[true]' to '--reverse') | |
| | polars sort-by count --reverse [true] | |
| # Fetch the top 5 pairs | |
| | polars first 5 | |
| # Convert back to a standard Nushell table to view | |
| | polars collect | |
| | polars into-nu | |
| ╭───┬───────────┬────────────────┬───────╮ | |
| │ # │ product_a │ product_b │ count │ | |
| ├───┼───────────┼────────────────┼───────┤ | |
| │ 0 │ Salsa │ Tortilla Chips │ 5358 │ | |
| │ 1 │ Bananas │ Milk │ 4483 │ | |
| │ 2 │ Bananas │ Bread │ 2970 │ | |
| │ 3 │ Bananas │ Salsa │ 2874 │ | |
| │ 4 │ Bananas │ Tortilla Chips │ 2857 │ | |
| ╰───┴───────────┴────────────────┴───────╯ | |
| #********************************************************************** | |
| # nushell with polars (0.5 seconds) | |
| # from luno.ai | |
| # Load, select relevant columns, deduplicate (transaction_id, product_name) pairs | |
| let df = ( | |
| polars open grocery_transactions.csv | |
| | polars select transaction_id product_name | |
| | polars drop-duplicates | |
| ) | |
| # Create left/right copies with renamed product columns for self-join | |
| let df_left = ($df | polars rename product_name prod_a) | |
| let df_right = ($df | polars rename product_name prod_b) | |
| # Self-join on transaction_id to generate all product combinations per transaction | |
| $df_left | |
| | polars join $df_right transaction_id transaction_id | |
| # Keep only unique unordered pairs (prod_a < prod_b eliminates self-pairs and dupes) | |
| | polars filter ((polars col prod_a) < (polars col prod_b)) | |
| # Count how many transactions contain each pair | |
| | polars group-by prod_a prod_b | |
| | polars agg [(polars len | polars as "count")] | |
| | polars collect | |
| # Sort by count descending and take top 5 | |
| | polars sort-by count -r [true] | |
| | polars first 5 | |
| ╭───┬─────────┬────────────────┬───────╮ | |
| │ # │ prod_a │ prod_b │ count │ | |
| ├───┼─────────┼────────────────┼───────┤ | |
| │ 0 │ Salsa │ Tortilla Chips │ 5358 │ | |
| │ 1 │ Bananas │ Milk │ 4483 │ | |
| │ 2 │ Bananas │ Bread │ 2970 │ | |
| │ 3 │ Bananas │ Salsa │ 2874 │ | |
| │ 4 │ Bananas │ Tortilla Chips │ 2857 │ | |
| ╰───┴─────────┴────────────────┴───────╯ | |
| #********************************************************************** | |
| # xan code (with nushell) (0.5 seconds) | |
| # from lumo.ai | |
| # Top 5 product pairs using Xan (final corrected version) | |
| try {rm temp.csv} | |
| # Step 1: Select and deduplicate | |
| xan select transaction_id,product_name grocery_transactions.csv | |
| | xan dedup -s transaction_id,product_name | |
| | save temp.csv | |
| # Step 2: Self-join, filter to unique unordered pairs, create pair column | |
| xan join -r right_ -D left transaction_id temp.csv temp.csv | |
| | xan filter "lt(product_name, right_product_name)" | |
| | xan map "concat(product_name, ' & ', right_product_name) as pair" | |
| # Step 3: Frequency already returns results sorted by count descending | |
| # frequency -s pair -l 5 gives us the top 5 pairs by count | |
| | xan frequency -s pair -l 5 | |
| # Step 4: Convert to Nushell table | |
| | from csv | |
| ## Cleanup | |
| #rm temp.csv | |
| ╭───┬───────┬──────────────────────────┬────────╮ | |
| │ # │ field │ value │ count │ | |
| ├───┼───────┼──────────────────────────┼────────┤ | |
| │ 0 │ pair │ Salsa & Tortilla Chips │ 5358 │ | |
| │ 1 │ pair │ Bananas & Milk │ 4483 │ | |
| │ 2 │ pair │ Bananas & Bread │ 2970 │ | |
| │ 3 │ pair │ Bananas & Salsa │ 2874 │ | |
| │ 4 │ pair │ Bananas & Tortilla Chips │ 2857 │ | |
| │ 5 │ pair │ <rest> │ 156773 │ | |
| ╰───┴───────┴──────────────────────────┴────────╯ | |