Everything FastStats can do
20 chart types, 31 statistical analyses, 27 data wrangling steps, an AI assistant, and import/export for the file formats you already use. Nothing here needs an upload, a Python environment or an account.
Import anything
Use the upload button, drag a file anywhere on the page, paste raw text, or point at a URL.
- CSV
- TSV
- Excel (.xlsx / .xls), with a picker for multi-sheet workbooks
- JSON
- NDJSON / JSON Lines
- GeoJSON
- Parquet
- Pasted text, right in the browser
- A URL pointing at any of the above
Export everywhere
Take your cleaned data, your pipeline, or your finished report with you — nothing is locked into FastStats.
- CSV, or copy straight to the clipboard
- JSON
- Excel (.xlsx)
- Parquet
- Markdown or standalone HTML report
- Reusable recipe file — replay a pipeline on new data
- Session file — data and pipeline together, reopenable later
20 chart types on Plotly.js
Point-and-click aesthetics, trend lines, log/linear scales, light and dark themes, seven palettes, and faceting into small multiples by any variable.
- Scatter
- Line
- Area
- Histogram
- Density
- ECDF
- Box
- Violin
- Strip / jitter
- Bar
- Pie / donut
- Pareto
- Treemap
- Sunburst
- 2D density
- Q-Q (normal)
- Heatmap (X × Y counts)
- Scatter matrix
- Correlation matrix
- Auto (picks a sensible default from your columns)
31 statistical analyses
From descriptive statistics through ANOVA, regression, PCA and clustering — every test runs client-side and writes its own plain-English interpretation.
- Descriptive statistics
- Normality (D’Agostino–Pearson)
- One-sample t-test
- Two-sample t-test (Welch)
- Paired t-test
- One-way ANOVA + Tukey HSD
- Two-way ANOVA (factorial)
- Kruskal-Wallis
- Mann-Whitney U
- Wilcoxon signed-rank
- Levene (equal variance)
- F-test (two variances)
- Pearson correlation
- Spearman correlation
- Kendall τ correlation
- Correlation matrix
- Outlier detection (IQR & z-score)
- Linear regression
- Multiple regression
- Logistic regression
- Chi-squared test
- Chi-squared goodness-of-fit
- Fisher exact (2×2)
- McNemar test (paired 2×2)
- One-proportion test
- Two-proportion z-test
- Bootstrap CI (mean / median)
- Cronbach’s α (scale reliability)
- PCA (principal components)
- K-means clustering
- Power / sample size (t-test)
27 data wrangling steps
Stack steps into a pipeline — filter, mutate, join or append a second file, group and summarize, pivot, window functions — and save the whole thing as a reusable recipe.
- Select columns
- Mutate column
- Rename column
- Filter rows
- Sort rows
- Keep first / last rows
- Sample rows
- Keep distinct combinations
- Drop missing rows
- Replace missing values
- Group rows
- Summarize
- Pivot longer
- Pivot wider
- Bin numeric column
- Extract date part
- Recode values
- Standardize (z / min-max)
- Clip outliers (winsorize)
- Split column
- Combine columns
- Fill missing (down / up)
- Window function (lag, cumsum…)
- Convert column type
- Join a second dataset
- Append rows from a second dataset
- Clean column names
An AI assistant that edits your project, not just your prompt
Describe a change in plain English — “filter to species setosa and plot petal length against width” — and the assistant edits your wrangling pipeline, chart, analysis or report directly. Every change is validated against your real column names and types before it's applied, and everything it does is undoable from the panel.
Only a digest of your project — column names, types, a few example values and one-line summaries of your chart, analysis and report — ever reaches the model. Your rows stay in your browser.
Try it on your own data
No account needed to start, and four sample datasets are loaded in if you want to look around first.