Control charts

Method

• Chart type, Auto: one value per row → I-MR; subgroups of up to 8 values → X̄-R; larger subgroups → X̄-S (the largest subgroup decides). Rows with one value each are charted as individual values.
• σ estimate (Minitab defaults): I-MR: MR̄ / d2(2) with moving ranges of length 2. X̄-R: Rᵢ / d2(nᵢ) over subgroups with at least 2 values, weighted by fᵢ = d2(nᵢ)² / d3(nᵢ)² (Minitab’s R̄ method), which is R̄ / d2 when all subgroups have the same size. X̄-S: sᵢ / c4(nᵢ) weighted by hᵢ = c4(nᵢ)² / (1 − c4(nᵢ)²), which is S̄ / c4 for equal sizes.
• Limits are 3σ of the plotted statistic. Mean chart: mean ± 3σ / √nᵢ, so unequal subgroups give limits that vary from subgroup to subgroup. R chart: d2σ ± 3 d3σ; S chart: c4σ ± 3σ√(1 − c4²); MR chart: d2(2)σ ± 3 d3(2)σ. A lower limit below 0 on a dispersion chart is set to 0.
• Center line of the mean chart is the mean of all measurements. Phase II uses the μ and σ you enter for both charts.
• Phase I and Phase II: in Phase I the limits are estimated from the data shown, so every new point also moves the limits; use it to judge whether a process was stable. In Phase II the limits come from a stable baseline and stay fixed, so new points are judged against them. Freeze limits takes μ and σ from the first N subgroups (25 by default) or from all data so far with the Phase I estimates above, switches to Phase II and fixes the chart type; the frozen values are saved in the settings file. Changing μ, σ, the chart type or going back to Phase I ends the freeze.
• Live monitoring (Chrome and Edge, File System Access API) reads a file that a test stand appends rows to, or a folder where every new file (matching a pattern such as *.csv) is one subgroup or adds its rows. Changes are found by the browser's change notifications (FileSystemObserver) where available, otherwise by checking every 1–60 s (5 s by default). A file is read only after its size and time stayed the same for two checks, so a half-written file is never read; every file read is remembered by name, time and size and never read twice (a file appended to adds only its new rows). Files go through the same parser, encodings and “Read as” choices as a paste. Rows found when watching starts are history and raise no alert; a new out-of-control signal afterwards shows a banner, “(!)” in the page title and, if chosen, a short sound.
• Monitoring limits: the page must stay open (a background tab may be checked less often by the browser), the browser asks for permission to read the file or folder, and files are only read, never changed. Sub-folders are not watched. In a folder, files are taken oldest first by their time. This offline file does not remember the choice: choose the file or folder again each time you open it. Nothing is sent anywhere.
• Tests follow the Nelson / Minitab numbering with Minitab's fixed parameters: 1 point beyond 3σ; 9 in a row on one side; 6 in a row rising or falling; 14 in a row alternating; 2 of 3 beyond 2σ (same side); 4 of 5 beyond 1σ (same side); 15 in a row within 1σ; 8 in a row beyond 1σ. Minitab's default is test 1 only; the others are switched on one by one. The range charts use tests 1–4 only.
• Flagged point: the point that completes the pattern, and every later point of the same run that still fits it. Runs and windows do not cross an empty row.
• Conclusion: “In control” when no enabled test fires, otherwise the number of signals (a point that fails two tests counts twice).
• Limits estimated from fewer than 20 points are unreliable, and control charts assume the measurements were taken in time order.
• Files: CSV, TSV and TXT are read as if pasted (UTF-8, UTF-16, otherwise Windows-1257). XLSX: choose the sheet and the columns or a cell range; formula cells use the value Excel last saved, dates are kept as text. Files up to 20 MB are read in this browser only and never uploaded.
• XLSX files are read with SheetJS Community Edition 0.20.3, Apache License 2.0 (license text at the end of this file).