Function analyzeTimeDurations

Analyzes an array of measured time durations and returns useful statistics both for the full dataset and for outlier-filtered data.

This is intended for benchmarking or performance-profiling scenarios where you collect many individual operation durations (e.g., using performance.now()).

What each metric means

• count: Number of measured durations.
• durations: Sorted list of all input durations (ascending).
• filtered: Sorted list of durations excluding statistical outliers using the IQR rule.
• filteredCount: Number of durations remaining after outlier removal.

For both full and filtered stats:

• min: The shortest measured duration.
• max: The longest measured duration.
• mean: The arithmetic average time.
• median: The middle value when sorted (50th percentile). More stable than mean.
• std: Standard deviation. Measures how much the durations vary.
• p95: 95th percentile. 95% of values are below this value. Useful as a near-worst-case indicator.
• q1: 25th percentile (lower quartile).
• q3: 75th percentile (upper quartile).
• iqr: Interquartile range (Q3 − Q1). Measures the spread of the middle 50% of values.

Outlier filtering (IQR rule)

Outliers are removed if they fall outside: [Q1 − 1.5×IQR, Q3 + 1.5×IQR]

This avoids skewing results when a single operation takes unusually long.

How to collect durations

const durations: number[] = [];
for (let i = 0; i < iterations; i++) {
  const start = performance.now();
  await doWork();
  durations.push(performance.now() - start);
}

Important:

• Do NOT use Date.now(); it has only millisecond resolution.
• Pre-warm your code (JIT, caches, I/O) before measuring.
• Run many iterations (100–10 000) to get stable stats.
• Avoid logging inside the measured code; it distorts timing.

Returns

An object containing sorted durations, filtered durations, and statistics for both full and filtered datasets.

Throws

If the input array is empty.