Calculay

Statistics Calculator

Enter a series of numbers separated by a comma or space to instantly discover its descriptive data profile, including measures of central tendency.

Descriptive Statistics

Enter numbers separated by commas or spaces.

Mean (Average)

18

Median

15.5000

Mode

None

Range

38

Sum

108

Count (n)

6

Min / Max

4 / 42

Geometric Mean

13.9655

Statistics Calculator: Descriptive Analysis, Central Tendency, Variance & Dispersion Formulas

Descriptive statistics converts unstructured, high-dimensional numerical observations into meaningful parametric summaries. In scientific research, clinical trials, macroeconomic modeling, and machine learning feature engineering, raw data must be characterized across two essential dimensions: central tendency (identifying the expected clustering centroid) and dispersion (quantifying variability and spread around that centroid). The Calculay Statistics Calculator processes numerical arrays to compute the arithmetic mean, median, mode, sample variance (s²), population variance (σ²), standard deviation, quartiles (Q_1, Q_2, Q_3), interquartile range (IQR), and skewness.

Master Descriptive Statistics Formula Reference

Statistical MetricMathematical FormulaSensitivity to OutliersPrimary Analytical Use Case
Arithmetic Mean (x̄)Σ xᵢ ÷ nExtremely SensitiveSymmetric, bell-curved normal distributions
Median (x̃)Middle element of sorted arrayRobust (Immune)Skewed distributions (household income, home sale prices)
Sample Variance (s²)Σ( xᵢ - x̄ )² ÷ ( n - 1 )SensitiveUnbiased estimator of unknown population parameter (Bessel's correction)
Population Variance (σ²)Σ( xᵢ - μ )² ÷ NSensitiveTrue parameter when complete census dataset is captured (N)
Interquartile Range (IQR)Q₃ - Q₁Highly RobustCaptures middle 50% spread; basis for Tukey boxplot outlier fences

Why Sample Variance Divides by (n - 1): Bessel's Correction

A frequent point of confusion in applied statistics is why sample variance divides by n - 1 instead of sample size n. When drawing a finite sample from a broader population, the sample mean x̄ will naturally cluster closer to the sample points than the true, unknown population mean μ would.

Dividing by n produces a systematically underestimated (biased) variance. Dividing by the degrees of freedom, n - 1, mathematically corrects for this bias, ensuring that the expected value of the sample variance E[s²] asymptotically equals the true population variance σ².

Step-by-Step Worked Statistical Analysis

Dataset (n = 8): { 12, 15, 18, 22, 22, 28, 30, 45 }

  1. Sum of observations: 12 + 15 + 18 + 22 + 22 + 28 + 30 + 45 = 192
  2. Arithmetic Mean (x̄): 192 ÷ 8 = 24.0
  3. Median: Even n, average of 4th and 5th elements (22 and 22) → 22.0
  4. Mode: Most frequent element → 22 (frequency = 2)
  5. Quartiles: Q_1 = (15 + 18)/2 = 16.5, Q_3 = (28 + 30)/2 = 29.0
  6. Interquartile Range: IQR = 29.0 - 16.5 = 12.5
  7. Sum of Squared Deviations: (12-24)² + (15-24)² + (18-24)² + (22-24)² + (22-24)² + (28-24)² + (30-24)² + (45-24)² = 144 + 81 + 36 + 4 + 4 + 16 + 36 + 441 = 762.0
  8. Sample Variance (s²): 762.0 ÷ (8 - 1) = 762.0 ÷ 7 = 108.86
  9. Sample Standard Deviation (s): √108.86 = 10.43
  10. Tukey Outlier High Fence: Q_3 + (1.5 × IQR) = 29.0 + (1.5 × 12.5) = 29.0 + 18.75 = 47.75 (Observation 45 is within normal boundary limits).

The Empirical Rule (68–95–99.7) for Gaussian Distributions

For any dataset that follows a standard Gaussian normal distribution curve:

  • μ ± 1σ: Contains approximately 68.27% of all individual data points.
  • μ ± 2σ: Contains approximately 95.45% of all data points.
  • μ ± 3σ: Contains approximately 99.73% of all data points. Any point situated beyond 3σ occurs less than 0.27% of the time, designating it as a statistically extreme outlier.