Chart a property's distribution#

Pick a property and read the chart it produces, plus the summary figures for the features you can currently see.

The statistics panel answers questions about a layer's data rather than its geography: how values are distributed, which categories dominate, whether two properties move together. It sits inside the attribute table behind a Statistics toggle, and it starts collapsed. If the layer has no properties, the panel does not appear at all.

It always works on what you can currently see. Geo Map applies every filter and the timeline first, so a narrower table gives narrower statistics.

Pick a property to chart#

Open the panel and it shows a grid of the layer's properties, each tagged with its type: num, cat, time, or id. Click one and the panel switches into its explorer view, with that property on the X axis. Click the same one again to clear it, or click the close button to go back to the property list.

In the explorer, X is the property you examine and Y is what you measure about it. Y defaults to (count), which means 'how many features'. That is what you want for a distribution. Choose a numeric property for Y instead to measure that property across the X groupings. A change to X resets Y back to (count).

The chart picks itself#

There is no chart type control. The pairing of X and Y decides the chart, on the principle that only one kind makes sense for a given pair of types.

XYYou get
Numeric(count)A histogram of 20 bins
Timestamp(count)A timeline of 30 buckets
Categorical(count)A bar chart of the top 20 categories
NumericNumericA scatter plot, with a correlation coefficient
CategoricalNumericA bar chart of the top 20 categories, aggregated
TimestampNumericA line across 30 time buckets, aggregated
CategoricalCategoricalA bar chart of the top 15 categories

Geo Map samples scatter plots down to 2000 points, so a very large layer still charts quickly.

Set the aggregation#

When Y is numeric and X groups features into buckets, an Agg control appears. It decides how Geo Map combines the values in each bucket: Mean, Sum, Count, Min, Max, or Median. It starts on Mean, and the chart's Y axis names whichever one you pick.

The summary figures#

A grid of statistics sits below the chart, and it changes with the type of the X property.

  • Numeric: gives Min, Max, Mean, Median, Std Dev, Count, and Null, plus Corr (r) when a scatter plot has produced a correlation.
  • Timestamp: gives Earliest, Latest, Span, Count, and Null.
  • Categorical: gives Unique, Top, Count, and Null.

The panel abbreviates large numbers, so 1,200,000 reads as 1.2M. A missing value shows a dash. When a filter is active, Count shows the filtered number against the total, which is the quickest way to confirm a filter does what you meant.