Your CSV must have at least two columns: a date column and a numeric value column. A few common issues to avoid:
Numeric values with comma separators (e.g. 1,234,567) — these can break the CSV structure if not properly quoted. Before saving as CSV from Excel, format your value column as a plain number with no thousands separator, or ensure values are wrapped in quotes. Run the Data Validator to check for this.
Merged cells in Excel — merged date cells leave blank rows in the CSV. Unmerge all cells and fill each row with its own date before saving.
Mixed date formats — ensure all dates in the date column use the same format throughout. The Data Validator will flag inconsistencies.
Unsupported date formats — financial year formats like "2004-05" are not supported. Convert to a proper date (e.g. 01/04/2004) before uploading.
Getting Started
Prepare your data as a CSV file with at least two columns: a date column and a numeric value column. Then drag and drop the file onto the upload area, or click to browse.
Settings
X-Axis (Date): Select the column containing your dates.
Y-Axis (Value): Select the column containing your numeric data.
Date Style: Choose UK (DD/MM/YYYY), US (MM/DD/YYYY), or Auto-detect. Auto-detect works for most files but set this manually if dates parse incorrectly.
Start / End Date: Optionally filter your data to a specific date range.
Turn Length: Minimum number of points required in a segment before a change point can be detected. Default 5. Increase for smoother results on noisy data.
Loops: Number of bootstrap resamples. Higher values give more stable confidence levels. 1,000 is fast; use 5,000–10,000 for governance reports.
Show 3SD Limits: Adds ±3 standard deviation bands to the Step Change chart.
Chart Types
Step Change Analysis: Bootstrap CUSUM — detects multiple sequential step-changes with confidence levels. Main governance chart.
Stage Summary: Table showing each stage with start date, end date, mean, SD, confidence level, and % change from previous stage.
X-mR Chart: Shewhart Individuals and Moving Range chart. Shows UNPL, LNPL, and URL control limits.
Run Chart: Plots data against the median. Highlights runs of 6+ consecutive points above or below the median.
CUSUM Chart: The raw cumulative sum of deviations from the overall mean.
Raw Data: The original data points with no analysis overlay.
Exporting
Use Export PNG to save the current chart as an image. Use Export PDF to generate a formatted report suitable for board papers and governance submissions. Both buttons appear after analysis is complete.
Data Validator
If your analysis produces unexpected results or an error, click Data Validator to run a diagnostic scan on your CSV. It checks for out-of-order dates, missing values, blank cells, and format errors.
Privacy
Your CSV file never leaves your computer. All analysis runs entirely in your browser. No data is uploaded to any server. No login required. Works offline once loaded.
About the method
Bootstrap CUSUM step-change analysis is a distribution-free method for detecting structural changes in time series data. It does not assume normality, making it particularly suited to healthcare safety data, incident counts, and other right-skewed metrics. The confidence level is earned from your data by resampling — not assumed from a statistical formula.
Based on Taylor (2000), building on Page (1954), Hinkley (1971), and Efron & Tibshirani (1993). Implemented at StepChangeAnalysis.com.