Statistical analysis utilities. Pure functions with zero dependencies - can be used standalone for any data analysis.
23 declarations
(values: number[], options?: StatsConfidenceIntervalOptions | undefined): [number, number] Calculate confidence interval for the mean.
values- Array of numbers
number[]options?- Configuration options
StatsConfidenceIntervalOptions | undefined[number, number] [lower_bound, upper_bound]
(mean: number, std_dev: number, sample_size: number, options?: StatsConfidenceIntervalOptions | undefined): [number, number] Calculate confidence interval from summary statistics (mean, std_dev, sample_size). Useful when raw data is not available.
mean- Mean of the data
numberstd_dev- Standard deviation of the data
numbersample_size- Number of samples
numberoptions?- Configuration options
StatsConfidenceIntervalOptions | undefined[number, number] [lower_bound, upper_bound]
(level: number): number Convert a confidence level (0-1) to a z-score. Uses a lookup table for common values, approximates others.
levelnumbernumber Record<number, number> Common z-scores for confidence intervals.
(mean: number, std_dev: number): number Calculate the coefficient of variation (CV). CV = standard deviation / mean, expressed as a ratio. Useful for comparing relative variability between datasets.
meannumberstd_devnumbernumber (x: number, a: number, b: number): number Approximate regularized incomplete beta function for p-value calculation. Uses continued fraction expansion for reasonable accuracy.
xnumberanumberbnumbernumber (z: number): number Log gamma function approximation (Lanczos approximation).
znumbernumber (values: number[]): number Calculate the mean (average) of an array of numbers.
valuesnumber[]number (values: number[]): number Calculate the median of an array of numbers.
valuesnumber[]number (values: number[]): { min: number; max: number; } Calculate min and max values.
valuesnumber[]{ min: number; max: number; } (x: number): number Standard normal CDF approximation (Abramowitz and Stegun formula 7.1.26).
xnumbernumber (values: number[], options?: StatsOutliersIqrOptions | undefined): StatsOutlierResult Detect outliers using the IQR (Interquartile Range) method. Values outside [Q1 - multiplier*IQR, Q3 + multiplier*IQR] are considered outliers.
valuesnumber[]options?StatsOutliersIqrOptions | undefinedStatsOutlierResult (values: number[], options?: StatsOutliersMadOptions | undefined): StatsOutlierResult Detect outliers using the MAD (Median Absolute Deviation) method. More robust than IQR for skewed distributions. Uses modified Z-score: |0.6745 * (x - median) / MAD| Values with modified Z-score > threshold are considered outliers.
valuesnumber[]options?StatsOutliersMadOptions | undefinedStatsOutlierResult (values: number[], p: number): number Calculate a percentile of an array of numbers using linear interpolation. Uses the "R-7" method (default in R, NumPy, Excel) which interpolates between data points for more accurate percentile estimates, especially with smaller samples.
values- Array of numbers
number[]p- Percentile (0-1, e.g., 0.95 for 95th percentile)
numbernumber (values: number[], mean?: number | undefined): number Calculate the standard deviation of an array of numbers. Uses population standard deviation (divides by n, not n-1). For benchmarks with many samples, this is typically appropriate.
valuesnumber[]mean?number | undefinednumber (t: number, df: number): number Approximate two-tailed p-value from t-distribution. For large df (>100), uses normal approximation. For smaller df, uses incomplete beta function.
t- Absolute value of t-statistic
numberdf- Degrees of freedom
numbernumber Two-tailed p-value
(values: number[], mean?: number | undefined): number Calculate the variance of an array of numbers.
valuesnumber[]mean?number | undefinednumber (mean1: number, std1: number, n1: number, mean2: number, std2: number, n2: number): StatsWelchTTestResult Calculate Welch's t-test statistic and degrees of freedom. Welch's t-test is more robust than Student's t-test when variances are unequal.
mean1- Mean of first sample
numberstd1- Standard deviation of first sample
numbern1- Size of first sample
numbermean2- Mean of second sample
numberstd2- Standard deviation of second sample
numbern2- Size of second sample
numberStatsWelchTTestResult StatsConfidenceIntervalOptions Configuration options for confidence interval calculation.
z_scoreZ-score for confidence level (default: 1.96 for 95% CI)
numberconfidence_levelConfidence level (0-1), alternative to z_score. If both provided, z_score takes precedence.
numberStatsOutlierResult Result from outlier detection.
cleanedValues after removing outliers
Array<number>outliersDetected outlier values
Array<number>StatsOutliersIqrOptions Configuration options for IQR outlier detection.
iqr_multiplierMultiplier for IQR bounds (default: 1.5)
numbermin_sample_sizeMinimum sample size to perform outlier detection (default: 3)
numberStatsOutliersMadOptions Configuration options for MAD outlier detection.
z_score_thresholdModified Z-score threshold for outlier detection (default: 3.5)
numberz_score_extremeExtreme Z-score threshold when too many outliers detected (default: 5.0)
numbermad_constantMAD constant for normal distribution (default: 0.6745)
numberoutlier_ratio_highRatio threshold to switch to extreme mode (default: 0.3)
numberoutlier_ratio_extremeRatio threshold to switch to keep-closest mode (default: 0.4)
numberoutlier_keep_ratioRatio of values to keep in keep-closest mode (default: 0.8)
numbermin_sample_sizeMinimum sample size to perform outlier detection (default: 3)
numberiqr_optionsOptions to pass to IQR fallback when MAD is zero
StatsWelchTTestResult Result from Welch's t-test calculation.
t_statisticThe t-statistic
numberdegrees_of_freedomWelch-Satterthwaite degrees of freedom
number