Module stats 

Column statistics (#261): count, nulls, distinct values, min, max, mean, median, quantiles and the most common values, per column.

A column is numeric when every non-null cell is a number, or text that parses as one (infer::parse_float); its min and max are numeric, and it gets a mean, a median and quantiles (linear interpolation between closest ranks, as NumPy’s default). Any other column’s min and max compare the cells’ text. Nulls are Value::Null and empty text; run Inference::apply first for other null tokens.

Stats renders as a table, a row per column; Stats::headers gives a short summary to show under each column heading instead.

use rich_data::stats::Stats;
use rich_data::{Rows, Value};

let mut rows = Rows::new(["service", "p99"]);
for (service, p99) in [("web", 120), ("api", 35), ("web", 80), ("db", 5)] {
    rows.push([service.into(), Value::Int(p99)]);
}
let stats = Stats::of(&rows);
let p99 = &stats.columns()[1];
assert_eq!((p99.count, p99.nulls, p99.distinct), (4, 0, 4));
assert_eq!(p99.mean, Some(60.0));
assert_eq!(p99.median, Some(57.5));
let service = &stats.columns()[0];
assert_eq!(service.top[0], ("web".to_string(), 2));
assert_eq!(service.min.as_deref(), Some("api"));

Structs

ColumnStats

One column’s statistics.

Stats

Statistics for every column. See the module docs.

StatsOptions

What Stats computes beyond the counts.

Functions

number

A number for display: whole numbers without a point, others to at most four decimals, and very small ones in scientific notation.