Interview Questions: Part 1
Statistics.
Data science and machine learning interviews almost always include a few statistics questions. Here are the most common ones, with short answers you could give in thirty seconds.

- Stats questions are common
- Short, clear answers
- Thirty seconds each
Mean or median? And what is a p-value?
When would you use the median instead of the mean? When the data is skewed, or has outliers, like incomes. What is a p value? The probability of seeing data at least this extreme, if the null hypothesis were true. It's not the probability that the null is true.
| key | value |
|---|---|
| Median | skewed data, outliers |
| p-value | P(data this extreme | null) |
| Not | P(null is true) |
| Small p | evidence against the null |
Errors and intervals
What's the difference between type one and type two errors? Type one is a false alarm, rejecting a true null. Type two is a miss, failing to detect a real effect. What does a ninety five percent confidence interval mean? If we repeated the study many times, ninety five percent of the intervals would contain the true value.
| key | value |
|---|---|
| Type I | false alarm, rate α |
| Type II | a miss, rate β |
| Power | 1 − β |
| 95% CI | 95% of intervals catch the truth |
Correlation and the CLT
Does correlation imply causation? No. A confounder can drive both variables, and only experiments can prove cause. What does the central limit theorem say? Sample means are approximately normal, whatever the original distribution, once the sample is large enough. That's why so many tests work.
| key | value |
|---|---|
| Correlation | not causation |
| Confounder | a hidden common cause |
| CLT | sample means become normal |
| Why it matters | tests and intervals |
Summary
Know the mean versus median, p values, the two error types, confidence intervals, and the central limit theorem. Next, interview questions on the math behind machine learning.
Key takeaways- p-value meaning
- Type I vs II
- CLT