Archive

2026

220 field notes published in 2026.

August 8

  1. The overall average hides 8 different numbers Analytics practice
  2. A essentially none correlation, and what it is not Statistics
  3. What 3,500 job postings say about city Analytics practice
  4. Correction: As-of feature computation Machine learning
  5. How much statistics do I actually need? Statistics
  6. Why there is no analytics on this site Data platform
  7. Charting revenue_usd by device Visualization
  8. 9.9%, plus or minus what? Statistics

July 31

  1. A five-minute use for Experiment Calculator Visualization
  2. SaaS subscriptions, in one paragraph Data quality
  3. Cutting a lesson down — Designing an experiment that can succeed Experimentation
  4. How the scheduled posts work Data engineering
  5. Pattern: Slowly changing dimension, type 2 Analytics engineering
  6. The time coverage of grid-energy-load Data engineering
  7. The category that eats the chart Visualization
  8. A running total that does not lie about ties Analytics engineering
  9. The total came out 8.2× too high Analytics engineering
  10. Counting rows is not testing grain Analytics engineering
  11. 57.8% of churned_at is missing Statistics
  12. 5,383 zeros in revenue_usd, and what they mean Data quality
  13. Before you clip the outliers Statistics
  14. The average distance km is not the typical one Statistics
  15. Reading reopened before trusting it in support-tickets Data quality
  16. Reading 3 columns instead of 12 Data platform
  17. gdp_per_capita_usd by region: a 105% spread Analytics practice
  18. r = 0.51 between distance_km and tip_usd Statistics
  19. Correction: Fan-out join Analytics engineering
  20. My experiment came back flat. Now what? Experimentation
  21. How scheduled posts work here Data platform
  22. Charting baseline_score by arm Visualization
  23. The interval is the result Statistics
  24. Grid energy load, in one paragraph Data quality
  25. What Data Explorer will not do Visualization
  26. Cutting a lesson down — The first hour with an unfamiliar dataset Data quality
  27. saas-subscriptions.plan: 4 values, 94.0% in the top three Visualization
  28. Zero is not a small number Data quality
  29. Date spine, in short Analytics engineering
  30. posted_on runs 2023-01-01 to 2025-09-27 Data engineering
  31. Comparing a period to the one before it Analytics engineering

June 30

  1. What wind_mw actually contains in grid-energy-load Data quality
  2. The total came out 4.2× too high Analytics engineering
  3. Is ticket_id + channel the grain of support-tickets? Analytics engineering
  4. 21.8% of salary_max_usd is missing Statistics
  5. The arithmetic behind a cheap query Data platform
  6. The average tip usd is not the typical one Statistics
  7. The top of bmi is 1.1× the 99th percentile Statistics
  8. revenue_usd by variant: a 48% spread Analytics practice
  9. r = 0.14 between load_mw and temp_c Statistics
  10. Correction: Gaps and islands Analytics engineering
  11. When should I not build a model? Machine learning
  12. The build takes under two seconds Data platform
  13. Charting flight_no by origin Visualization
  14. 3.0%, plus or minus what? Statistics
  15. If we had to cut the course to ten lessons Experimentation
  16. Why Chart Builder works the way it does Visualization
  17. Data job postings, in one paragraph Data quality
  18. Reader question — Reading a result without fooling yourself Experimentation
  19. Deduplicate by rank, in short Analytics engineering
  20. The time coverage of sensor-telemetry Data engineering
  21. flight-delays.dest: 14 values, 22.3% in the top three Visualization
  22. Is posting_id + country the grain of data-job-postings? Analytics engineering
  23. A running total that does not lie about ties Analytics engineering
  24. The total came out 1.0× too high Analytics engineering
  25. What device actually contains in ab-test-checkout Data quality
  26. 21.8% of salary_min_usd is missing Statistics
  27. arr_delay_min: mean 7, median 2 Statistics
  28. Zero is not a small number Data quality
  29. Reading 3 columns instead of 8 Data platform
  30. Before you clip the outliers Statistics

May 31

  1. dep_delay_min by carrier: a 85% spread Analytics practice
  2. A essentially none correlation, and what it is not Statistics
  3. Correction: Idempotent window replace Data engineering
  4. How do I explain a confidence interval to a stakeholder? Statistics
  5. Why the datasets are seeded Data platform
  6. Charting humidity_pct by status Visualization
  7. Zero is not a small number Data quality
  8. 42.1%, plus or minus what? Statistics
  9. SQL Playground works offline Visualization
  10. What is deliberately wrong with city-air-quality Data quality
  11. Five minutes with distance_mi in flight-delays Data quality
  12. Reader question — Turning a request into a question Analytics practice
  13. Pattern: Deduplicate by rank Analytics engineering
  14. hour_at runs 2023-01-01 to 2023-12-31 Data engineering
  15. dep_delay_min by origin: a 113% spread Analytics practice
  16. Check the blanks before you drop them Statistics
  17. device_id is not unique, and the join knows Analytics engineering
  18. ab-test-checkout.variant: 3 values, 100.0% in the top three Visualization
  19. A essentially none correlation, and what it is not Statistics
  20. Somebody translated the course into Portuguese Analytics practice
  21. first_response_min: mean 155, median 83 Statistics
  22. Is reading_at + device_id the grain of sensor-telemetry? Analytics engineering
  23. Sessions out of raw events in retail-orders Analytics engineering
  24. Charting sched_dep_hour by origin Visualization
  25. The arithmetic behind a cheap query Data platform
  26. The top of rpm is 1.1× the 99th percentile Statistics
  27. Correction: Deterministic surrogate key Data engineering
  28. The interval is the result Statistics
  29. Is a data engineering certificate worth it? Data engineering
  30. Content is files, not a database Data platform
  31. A five-minute use for Data Explorer Visualization

April 30

  1. What is deliberately wrong with flight-delays Data quality
  2. Course note — Colour is an encoding, not decoration Visualization
  3. Pattern: Baseline before model Machine learning
  4. signed_up_on runs 2023-01-04 to 2025-12-01 Data engineering
  5. The category that eats the chart Visualization
  6. Is subject_id + arm the grain of clinical-trial? Analytics engineering
  7. Finding the gaps in ab-test-checkout Analytics engineering
  8. user_id is not unique, and the join knows Analytics engineering
  9. 183 zeros in arr_delay_min, and what they mean Data quality
  10. Check the blanks before you drop them Statistics
  11. fare_usd: mean 22.16, median 14.29 Statistics
  12. Five minutes with year in world-indicators Data quality
  13. The arithmetic behind a cheap query Data platform
  14. Before you clip the outliers Statistics
  15. The overall average hides 4 different numbers Analytics practice
  16. A strong correlation, and what it is not Statistics
  17. One question I ask in every code review Analytics engineering
  18. Correction: Contract check at the boundary Data quality
  19. Is my pipeline actually idempotent? Data engineering
  20. Charting sched_dep_hour by carrier Visualization
  21. What "free" means here, precisely Data platform
  22. The interval is the result Statistics
  23. The chart rules behind Data Explorer Visualization
  24. Flight delays, in one paragraph Data quality
  25. Course note — The patterns that keep coming up Analytics engineering
  26. The category that eats the chart Visualization
  27. Pattern: Conditional pivot Analytics engineering
  28. The time coverage of saas-subscriptions Data engineering
  29. Comparing a period to the one before it Analytics engineering
  30. Zero is not a small number Data quality

March 31

  1. The total came out 34.6× too high Analytics engineering
  2. Counting rows is not testing grain Analytics engineering
  3. The overall average hides 6 different numbers Analytics practice
  4. The average sessions is not the typical one Statistics
  5. 1.6% of pm25 is missing Statistics
  6. Five minutes with cancelled in flight-delays Data quality
  7. Charting units by country Visualization
  8. The arithmetic behind a cheap query Data platform
  9. Before you clip the outliers Statistics
  10. r = 0.50 between temp_c and solar_mw Statistics
  11. The interval is the result Statistics
  12. Correction: Forward-chaining split Machine learning
  13. Why SQL Playground works the way it does Visualization
  14. What is deliberately wrong with saas-subscriptions Data quality
  15. Site search without a server, in about 80 lines Visualization
  16. retail-orders.category: 6 values, 59.5% in the top three Visualization
  17. Course note — A working setup that will not embarrass you Data engineering
  18. Merge with a recency guard, in short Data engineering
  19. pickup_at runs 2023-01-01 to 2023-12-31 Data engineering
  20. Keeping one row per key in retail-orders Analytics engineering
  21. Is resolved_at + channel the grain of support-tickets? Analytics engineering
  22. Check the blanks before you drop them Statistics
  23. gdp_per_capita_usd: mean 26,959, median 20,761 Statistics
  24. What salary_max_usd actually contains in data-job-postings Data quality
  25. Reading 3 columns instead of 10 Data platform
  26. Before you clip the outliers Statistics
  27. first_response_min by channel: a 11% spread Analytics practice
  28. A essentially none correlation, and what it is not Statistics
  29. Correction: The grain uniqueness test Data quality
  30. Charting salary_max_usd by seniority Visualization
  31. The interval is the result Statistics

February 28

  1. A five-minute use for Distribution Lab Visualization
  2. What is deliberately wrong with sensor-telemetry Data quality
  3. Course note — Joins that do not fan out Analytics engineering
  4. Slowly changing dimension, type 2, in short Analytics engineering
  5. The time coverage of ab-test-checkout Data engineering
  6. Check the blanks before you drop them Statistics
  7. The category that eats the chart Visualization
  8. Finding the gaps in saas-subscriptions Analytics engineering
  9. Is user_id + genre the grain of movie-ratings? Analytics engineering
  10. The average arr delay min is not the typical one Statistics
  11. Reading borough before trusting it in ride-hail-trips Data quality
  12. Reading 3 columns instead of 11 Data platform
  13. The top of revenue_usd is 2.8× the 99th percentile Statistics
  14. population by country: a 4,739% spread Analytics practice
  15. A weak correlation, and what it is not Statistics
  16. Correction: Sessionisation Analytics engineering
  17. Charting unit_price_usd by channel Visualization
  18. 2.1%, plus or minus what? Statistics
  19. This whole site is 900 lines of Node and one dependency Data engineering
  20. A five-minute use for Chart Builder Visualization
  21. Deliberately messy customers, in one paragraph Data quality
  22. Reader question — When not to build a model Machine learning
  23. ride-hail-trips.borough: 5 values, 73.6% in the top three Visualization
  24. Idempotent window replace, in short Data engineering
  25. The time coverage of city-air-quality Data engineering
  26. units: mean 2, median 1 Statistics
  27. A running total that does not lie about ties Analytics engineering
  28. Is device_id + status the grain of sensor-telemetry? Analytics engineering

January 31

  1. 2.1% of arr_delay_min is missing Statistics
  2. What ordered_at actually contains in retail-orders Data quality
  3. The arithmetic behind a cheap query Data platform
  4. The top of temp_c is 1.1× the 99th percentile Statistics
  5. The overall average hides 30 different numbers Analytics practice
  6. r = 0.80 between o3_ppb and temp_c Statistics
  7. Correction: Slowly changing dimension, type 2 Analytics engineering
  8. Charting csat by channel Visualization
  9. A five-minute use for Regression Lab Visualization
  10. What is deliberately wrong with retail-orders Data quality
  11. Cutting a lesson down — Owning something that runs without you Data platform
  12. reading_at runs 2023-01-01 to 2023-02-28 Data engineering
  13. Pattern: Deterministic surrogate key Data engineering
  14. The category that eats the chart Visualization
  15. revenue_usd: mean 111, median 60 Statistics
  16. Keeping one row per key in sensor-telemetry Analytics engineering
  17. Is ticket_id + queue the grain of support-tickets? Analytics engineering
  18. Six things we got wrong, corrected in place Analytics engineering
  19. Check the blanks before you drop them Statistics
  20. What unit_price_usd actually contains in retail-orders Data quality
  21. The arithmetic behind a cheap query Data platform
  22. The top of arr_delay_min is 3.1× the 99th percentile Statistics
  23. The overall average hides 5 different numbers Analytics practice
  24. A essentially none correlation, and what it is not Statistics
  25. Correction: Deduplicate by rank Analytics engineering
  26. Charting fare_usd by payment_type Visualization
  27. A five-minute use for Correlation Explorer Visualization
  28. An exercise on clinical-trial Data quality
  29. Course note — The five tests that cover most questions Experimentation
  30. Forward-chaining split, in short Machine learning
  31. flight_date runs 2023-01-01 to 2023-12-31 Data engineering

Back to the blog