What this calculator does
Descriptive statistics summarise a set of numbers in a few figures: where the centre is, how spread out the values are, what shape the distribution has, and whether any values stand apart from the rest. They are the first thing to check before any test or model, and the first thing an interviewer will ask you to compute from a dataset.
Paste one column of numbers and the calculator returns more than twenty statistics, each next to the Excel formula that reproduces it, plus a box plot, a histogram, a frequency table and a percentile table.
The statistics, grouped
- Centre: mean (the average), median (the middle value once sorted) and mode (the most frequent value). The geometric mean is included for growth rates and ratios when every value is positive.
- Spread: range, variance, standard deviation, interquartile range (IQR), coefficient of variation (standard deviation as a share of the mean) and the standard error of the mean.
- Position: minimum, Q1, median, Q3 and maximum, which together form the five-number summary, plus the 5th to 95th percentiles.
- Shape: skewness and excess kurtosis.
- Inference: a confidence interval for the population mean, using the t distribution.
Key formulas
IQR = Q3 − Q1 · outlier if x < Q1 − 1.5·IQR or x > Q3 + 1.5·IQR
SE = s / √n · CI = x̄ ± t* × SE
For a population, the variance divides by n instead of n − 1. Skewness and kurtosis use the
bias-adjusted sample formulas that Excel’s SKEW and KURT and
pandas use, or the plain moment formulas when you choose population.
Worked example: quick-commerce delivery times
Press Load example to load the delivery times, in minutes, of 25 quick-commerce orders in Bengaluru. Most arrived in 10 to 20 minutes, but two orders placed during heavy rain took 38 and 44 minutes.
- Mean 17.64 vs median 15. The two slow orders pull the mean up by more than two and a half minutes. “Typical delivery takes 15 minutes” is the more honest headline.
- Standard deviation 8.36 minutes and a coefficient of variation of 47%, so delivery times vary a lot relative to their average.
- Q1 = 12, Q3 = 19, IQR = 7. The upper fence is 19 + 1.5 × 7 = 29.5, so 38 and 44 are flagged as outliers.
- Skewness 1.97: a long right tail, as you would expect for times that cannot go below zero but can stretch out on a bad day.
- 95% CI for the mean: 14.19 to 21.09 minutes. Wide, because the sample is small and spread out.
Switch the quartile method to Exclusive and Q3 moves from 19 to 20. Neither is wrong; they are two conventions. Say which one you used when you report quartiles, and match the method your course or company uses.
Getting the same result in Excel and Python
The All statistics tab lists the Excel formula for each row. In Python, pandas gives most of them in one line (pandas uses the sample formulas and the inclusive quartile method by default):
import pandas as pd
minutes = pd.Series([9, 10, 11, 11, 12, 12, 12, 13, 13, 14, 14, 15, 15,
16, 16, 17, 18, 18, 19, 21, 22, 24, 27, 38, 44])
print(minutes.describe()) # count, mean, std, min, 25%, 50%, 75%, max
print(minutes.skew(), minutes.kurt()) # 1.9717, 4.0697
print(minutes.mode().tolist()) # [12]
Common mistakes
- Using population formulas on a sample. Dividing by n instead of n − 1 understates the spread, most noticeably for small samples.
- Reporting only the mean for skewed data. Salaries, order values and waiting times almost always need the median alongside the mean.
- Deleting outliers by reflex. The rain-day deliveries above are real and matter to customers. Investigate an outlier before deciding whether it is an error or the most important point in the data.
- Comparing standard deviations across different scales. A standard deviation of ₹500 means something different for ₹2,000 orders than for ₹50,000 orders. Use the coefficient of variation to compare.