Overview
Descriptive statistics summarise the observed dataset without testing a causal model. Survlytics reports centre, spread, shape, ranges, quartiles and mean confidence intervals for finite numeric observations.
Theoretical foundation
The arithmetic mean and sample moments describe the empirical distribution. Sample variance and standard deviation use the n − 1 denominator. Bias-corrected sample skewness and Fisher excess kurtosis are provided only when the sample varies and has enough observations.
Key equations
\bar{x}=\frac{1}{n}\sum_{i=1}^{n}x_is^2=\frac{1}{n-1}\sum_{i=1}^{n}(x_i-\bar{x})^2Algorithm and computation
The analysis service coerces selected columns to numeric, converts infinities to missing, and then uses NumPy 2.4.2 and SciPy 1.17.0. Quantiles use NumPy quantile. Confidence intervals for the mean use a Student t critical value with n − 1 degrees of freedom. Histograms use at most 20 bins and a square-root sample-size rule. Listwise mode first removes rows incomplete on any selected variable; available mode summarises each variable from its observed values.
Required data
One or more numeric variables. Ordinal variables require an explicit acknowledgement before interval-like summaries are calculated.
Assumptions
The confidence interval concerns the mean under an independent-sampling model. Skewness and kurtosis are descriptive sample estimates, not automatic normality decisions.
Available settings
Missing-data handling; confidence level of 90%, 95% or 99%; display precision; ordinal-treatment acknowledgement.
Worked example
Synthetic example: for five scores 2, 3, 3, 4 and 5, Survlytics reports a mean of 3.4 and uses the sample standard deviation with n − 1. This example is illustrative and is not a research finding.
Limitations and non-claims
No survey weights, complex-sample variance estimation or robust location estimators are implemented. A confidence interval does not correct biased sampling.
Validation notes
Covered by repository unit tests against independently calculated NumPy/SciPy expectations. No external package benchmark fixture is claimed for the complete descriptive table.
How Surv.lytics computes Descriptive statistics
{'confidence': 0.95, 'ddof': 1}Was this article helpful?
Your response helps the methodology team improve this documentation.