Is Your Data Ready for Analysis? Normality & Outlier Screening

How it works

For every column you map, the analysis computes the mean, median, standard deviation, skewness, and excess kurtosis, then runs the Shapiro-Wilk test of normality (on a random sample of 5,000 values when a column is larger than that). It counts outliers two ways — points beyond 1.5 times the interquartile range (mild) and beyond 3 times (extreme) — and runs a Grubbs test on the single most extreme value. Each column ends with a verdict and a recommendation: if the normality test rejects or the skewness exceeds one, a nonparametric method or a transformation is suggested; otherwise parametric methods are appropriate.

Use it right before a t-test, ANOVA, correlation, or regression, when those methods assume roughly normal, outlier-free data and you want to confirm each column qualifies.

Not a substitute for looking at the data — a histogram still reveals shape the summary numbers compress. It also screens columns one at a time and does not test relationships between them.

What data do you need?

Any table with numeric columns you plan to test. For example, experiment measurements:

Minimum 10 rows · Best with 30-5,000 rows and 1-12 numeric columns

What's in the report?

Standard-library analysis: before you run a t-test, ANOVA, or regression, check whether each numeric column is fit for those methods. For every column it reports the mean, median, spread, skewness and kurtosis, runs the Shapiro-Wilk normality test, counts outliers with the 1.5 and 3 times IQR rules plus a Grubbs test on the single most extreme value, and gives a clear parametric-versus-nonparametric recommendation. Works on any dataset: map the numeric columns you plan to test.