Custom visual

Statistics Suite Coming to AppSource

Five statistical charts Power BI doesn’t ship — box plots, violins, beeswarm, histogram and a correlation matrix — in one visual, with statistics you can defend in a meeting.

Get in touch See the features Limitations

Four box plots for the Lead, Senior, Mid and Junior salary bands with every employee drawn as a point around each box, a mean marker, sample sizes under each group and the four most extreme outliers named. — Box plots with every observation drawn on top, a mean marker, sample sizes, and the most extreme outliers named — one box per salary band.

Why this exists

Microsoft removed its own Histogram visual in 2022, along with the Percentile Chart. There is no correlation matrix on AppSource at all, because the only one there ever was is an R visual — and R visuals can never be certified. The box plots that remain are largely unmaintained, and several quietly sample your data at 30,000 rows without telling you. This visual replaces all of them.

Features

●Box plots that name their method. Quartiles by Tukey hinges, Excel’s QUARTILE.INC, or QUARTILE.EXC; whiskers by k×IQR, min–max, percentiles or ±k·SD. The active method is named in the tooltip, so nobody has to guess why your numbers differ from someone else’s.
●Validated against Excel. QUARTILE.INC, QUARTILE.EXC, MEDIAN, AVERAGE, STDEV.S and CORREL all match to the precision shown — plus Spearman against RANK.AVG.
●Real violins. Genuine kernel density estimation (Epanechnikov or Gaussian) with Silverman’s bandwidth rule and a manual override — not a smoothed histogram pretending to be a density.
●Every point, clickable. Beeswarm or deterministic jitter, all points or outliers only. Clicking a point cross-filters the rest of your report, so an outlier becomes a question you can answer: which employee, which order, which batch.
●Histogram. Freedman–Diaconis or Sturges bins, or set the count or width yourself. Count, percent or density normalization, with KDE and fitted-normal overlays that stay scale-correct in all three. Cumulative and Pareto lines, reference lines at mean, median, ±kσ or any values you type.
●Correlation matrix. Pearson or Spearman, computed natively with two-tailed p-values, significance stars, pairwise complete-case handling and a diverging colour scale.
●No silent sampling. Free box-plot visuals stop at 30,000 rows. R and Python visuals are hard-capped at 150,000. Statistics Suite loads to the Power BI platform maximum — and when the platform refuses more data it says so in a banner on the chart, instead of quietly showing you a wrong median. Tested to 600,000 rows.
●A format pane that stays out of the way. Fourteen small cards that show and hide with the chart type, so you never scroll past violin settings to find histogram bins.

Salary by department drawn as violins with a box plot and every employee as a point inside each; the violins show several bumps, one per salary band. — Violin, box and points are independent layers — use any combination. The curve is real kernel density, with bandwidth you control, and it shows what a box hides: each department has one bump per salary band.

A histogram of salary in ten round bins of 20,000 with counts on the bars, a fitted normal curve overlaid, and reference lines at the mean and at plus two standard deviations. — Histogram with round bins, a fitted normal overlaid and reference lines at the mean and +2σ. Curves stay correctly scaled whether the axis shows counts, percentages or density; a kernel density curve and Freedman–Diaconis bins are one setting away.

A three-by-three Pearson correlation matrix of Salary, Bonus and Tenure with significance stars, beside a scatter chart of Bonus against Salary showing what r = 0.97 looks like. — The chart AppSource has been missing entirely. Significance stars on the cells, and r, n and a two-tailed p-value in every tooltip. The scatter beside it is a native chart.

A box plot with two named outliers selected and every other point dimmed, next to a table visual filtered to those two employees. — An outlier isn’t a red dot — it’s a person, an order or a batch you can click through to. Here two clicked points filter the table beside it.

Where it fits

Strongest in HR compensation review — box plot, points, and click an outlier to see who. Also process capability, lab and quality data, survey analysis, and any “is this difference real?” question.

Honest limitations

We would rather you know these before you buy than after.

Distribution and histogram use the first Values field. Extra fields feed the correlation matrix. Add two or more for correlation mode.
The legend is ignored in histogram mode. Split violins, notches and confidence intervals, and small multiples are planned, not shipped.
Column format strings aren’t applied yet to axis and label numbers.
Above 50,000 values per group the violin curve is integrated over a quantile-spaced sample. Bandwidth and every reported statistic still use every row — only the drawn curve is approximated.

Learn it in fifteen minutes

The Statistics Suite tutorial walks through the three charts with a sample dataset you can download: how to read a box and make it agree with Excel, when a violin helps, outliers you can click, histogram binning, and correlation with p-values.

Questions about Statistics Suite?

Pricing, a feature you need, whether it fits your model — ask, and a person who wrote it answers within two business days. It is in AppSource certification now.

Get in touch Read the tutorial