Statistics for data analysis: 12 concepts every analyst uses
Data analysts mainly use descriptive statistics (mean, median, standard deviation, percentiles), distributions, sampling, confidence intervals, hypothesis tests and p-values, A/B testing, correlation and simple regression. You do not need advanced maths: you need to know which measure fits the data and how to explain what a result means.
1. Mean, median and mode
The mean is the average, the median is the middle value, and the mode is the most common value. Use the median for skewed data such as salaries or order values, because a few very large values pull the mean up.
2. Range, variance and standard deviation
These measure spread. Standard deviation is the typical distance of values from the mean, in the same units as the data. Two products can have the same average rating and very different standard deviations.
3. Percentiles and the interquartile range
The 90th percentile is the value below which 90% of observations fall. The interquartile range (IQR) is the spread of the middle 50% (Q3 minus Q1), and is the basis of box plots.
4. Outliers
A common rule flags values below Q1 − 1.5 × IQR or above Q3 + 1.5 × IQR. Investigate outliers before removing them: they are often data errors, but sometimes they are the most important customers.
5. Distributions and the normal curve
A distribution shows how often each value occurs. In a normal distribution about 68% of values fall within one standard deviation of the mean and about 95% within two. Many business metrics are skewed instead, so always plot a histogram first.
6. Population and sample
The population is everyone you care about; a sample is the part you measured. Conclusions are only as good as the sample, so watch for samples that leave out a group, such as surveying only app users about a website.
7. The central limit theorem
Averages of large enough random samples are approximately normally distributed, even when the underlying data is not. This is why confidence intervals and many tests work on business data.
8. Confidence intervals
A 95% confidence interval is a range built by a method that captures the true value in 95% of samples. "Average order value is ₹540, 95% CI ₹510 to ₹570" is more honest than "₹540".
9. Hypothesis testing and p-values
A hypothesis test asks whether a difference is larger than random chance would produce. The p-value is the probability of seeing a result at least this extreme if there were really no difference. A p-value below 0.05 is the common cut-off for "statistically significant", but it does not tell you whether the difference is big enough to matter.
10. A/B testing
An A/B test randomly splits users between two versions and compares a metric such as conversion rate. Decide the metric and sample size before starting, run it for full weeks, and do not stop early the moment the result looks significant.
11. Correlation is not causation
Correlation (from −1 to +1) measures how closely two variables move together. Ice-cream sales and sunburn are correlated because both rise in summer, not because one causes the other. Only a controlled experiment such as an A/B test shows cause.
12. Simple linear regression
Regression fits a line that predicts one variable from another, for example sales from ad spend. The slope tells you how much the outcome changes per unit of input, and R² tells you how much of the variation the line explains.
Frequently asked questions
How much statistics does a data analyst need?
When should I use the median instead of the mean?
What does a p-value of 0.05 mean?
Learn it from handwritten notes
Data Analytics: The Handwritten Notes: 116 illustrated pages, 26 chapters, drawn diagrams and worked examples. ₹199 in India, US$9 elsewhere.
See what is inside