Data Scientist
Statistics and probability questions
152 questions on Statistics and probability for data scientists, easiest first.
Easy 48
- Typical income in a skewed dataseteasy
- What standard deviation tells youeasy
- Positive test for a rare diseaseeasy
- Expected profit of a dice gameeasy
- Probability of A or Beasy
- Values within two standard deviationseasy
- Median of six numberseasy
- Z-score of an exam resulteasy
- Units of varianceeasy
- Exactly two heads in three flipseasy
- A correlation of -0.9easy
- Reading a regression coefficienteasy
- Survey sent only to app userseasy
- What the central limit theorem sayseasy
- Comparing beta users with everyone elseeasy
- Percentage change between dayseasy
- Grouping sales by montheasy
- Two independent servers both upeasy
- Churn rate among mobile userseasy
- Expected value of a raffle ticketeasy
- Converting units and the SDeasy
- Expected sign-ups from 100 visitseasy
- Arrivals in ten minuteseasy
- Beyond 1.96 standard deviationseasy
- Red after five blackseasy
- Choosing a panel of threeeasy
- Order of three speakerseasy
- A negative covarianceeasy
- Reading a P90 latencyeasy
- Average of a 0/1 columneasy
- Sum of seven with two diceeasy
- Ten heads in a roweasy
- Typical value of a categoryeasy
- Writing the null hypothesiseasy
- A real effect that was missedeasy
- z-test or t-testeasy
- Moving from 95% to 99% confidenceeasy
- SD and standard erroreasy
- What R² of 0.64 meanseasy
- Predicting far outside the dataeasy
- R² from a correlationeasy
- Parts of a time serieseasy
- A simple baseline for weekly patternseasy
- Uncertainty further into the futureeasy
- Making sure every emirate is coveredeasy
- Only happy customers replyeasy
- One point creates a correlationeasy
- Effect size in standard deviationseasy
Medium 59
- Reading a 95% confidence intervalmedium
- Overall rate flips when splitmedium
- Predicting a heavily skewed targetmedium
- Studying only active usersmedium
- Validating a forecasting modelmedium
- Making a trending series stationarymedium
- Halving a confidence intervalmedium
- resample raises a TypeErrormedium
- A 7-day window on irregular datesmedium
- Average price weighted by quantitymedium
- Days since the previous purchasemedium
- Row percentages in a crosstabmedium
- At least one six in four rollsmedium
- Exclusive events and independencemedium
- Shared birthdays in a small teammedium
- Spread of a differencemedium
- At least one faulty unit in a boxmedium
- A minute with no callsmedium
- Calls until the first salemedium
- Height at the 84th percentilemedium
- Transaction amounts with a long tailmedium
- Two defective items without replacementmedium
- Correlation after a currency changemedium
- Which branch has steadier salesmedium
- Switching doors in a game showmedium
- Are two events independentmedium
- When a normal approximation worksmedium
- When a yes/no metric is noisiestmedium
- Is phone insurance worth itmedium
- Average gap between arrivalsmedium
- Two servers in the same rackmedium
- Reading a Q-Q plotmedium
- Two branches combinedmedium
- Converting probability to oddsmedium
- Choosing a one-sided testmedium
- Before and after on the same usersmedium
- Interval for a survey sharemedium
- Are orders spread evenly over weekdaysmedium
- Comparing average baskets in four citiesmedium
- Small, very skewed samplesmedium
- Significant but not importantmedium
- An interval that includes zeromedium
- Building a bootstrap intervalmedium
- Sample variance of three valuesmedium
- R² keeps rising with more columnsmedium
- A U-shaped residual plotmedium
- Reading a city dummy coefficientmedium
- Which predictor matters moremedium
- Ridge versus lassomedium
- A coefficient with p = 0.30medium
- Slope from correlation and spreadsmedium
- Autocorrelation spike at lag 7medium
- MAPE on slow-moving itemsmedium
- Ramadan shifts every yearmedium
- Flagging an unusual daymedium
- Sampling whole schoolsmedium
- Capping extreme order valuesmedium
- City averages and individual peoplemedium
- A curved but steady relationshipmedium
Hard 45
- Coaching the worst storeshard
- Coefficients that keep flippinghard
- Click-through rate standard errorshard
- A link that only appears in a subsethard
- Error spread grows with the predictionhard
- Direction of omitted variable biashard
- Precision when prevalence dropshard
- Noisy feature, flatter slopehard
- Back-transforming log predictionshard
- Effect with an interaction termhard
- Rolling mean at the edgeshard
- Average lifetime with active customershard
- Great backtest, poor live forecasthard
- When the bootstrap breaks downhard
- Day-over-day change for many storeshard
- First quartile of four valueshard
- Different standard deviations for the same datahard
- Latest exchange rate before each tradehard
- Month-end dates with date_rangehard
- Already waited ten minuteshard
- Variance of a uniform waithard
- A sample mean that keeps jumpinghard
- Which plant made the faulty parthard
- Guarantee without normalityhard
- Expected number of correct matcheshard
- Count data with variance far above the meanhard
- Spread of a day's totalhard
- Making a simulation ten times more precisehard
- A density value of 4hard
- Two positive screening testshard
- Sampling most of a small populationhard
- Sample size for a ±3% marginhard
- Overlapping confidence intervalshard
- How a permutation test workshard
- Reading a Bayesian credible intervalhard
- Maximum likelihood for a click ratehard
- Interval for one new storehard
- A logistic coefficient of 0.693hard
- One store moves the whole linehard
- Reading a log-log coefficienthard
- Updating a smoothed forecasthard
- Forecasting a random walkhard
- Two trending series that correlatehard
- Choosing the AR orderhard
- Zero failures in 600 transactionshard