Distributions, inference, A/B testing, regression basics
Practise these as flashcards. One or two lines each; the ones you miss come back first.
Start freeThe mean is the average of all values, while the median is the middle value when data is sorted. Median is less sensitive to outliers than mean.
The normal distribution is symmetric and bell-shaped, defined by its mean and standard deviation. Most values cluster around the mean, and about 68% fall within one standard deviation.
A p-value is the probability of observing data as extreme as the sample, assuming the null hypothesis is true. A low p-value suggests evidence against the null hypothesis.
A confidence interval gives a range of values likely to contain the true population parameter, based on the sample estimate and a chosen confidence level.
Conditional probability is the likelihood of an event given another event has occurred. For example, if 30% of emails are spam and 80% of spam emails contain 'free', the probability an email contains 'free' given it's spam is 0.8.
The central limit theorem states that the distribution of sample means approaches normality as sample size increases, regardless of the population's distribution. This allows us to use normal-based confidence intervals and hypothesis tests even for non-normal data.
Type I error is rejecting the null hypothesis when it's actually true (false positive), while Type II error is failing to reject the null hypothesis when it's false (false negative). Balancing these is important for reliable testing.
Peeking, or checking results before the test is complete, increases the risk of false positives because the statistical assumptions are violated. It can lead to incorrect conclusions and wasted resources.
A linear regression coefficient represents the expected change in the dependent variable for a one-unit increase in the predictor, holding other variables constant. It quantifies the strength and direction of the relationship.
Correlation means two variables move together, but it doesn't imply one causes the other. Causation means changes in one variable directly result in changes in another, which requires controlled experiments or strong evidence.
Power is the probability that a test correctly rejects the null hypothesis when there is a true effect. High power reduces the risk of Type II errors and is influenced by sample size and effect size.
Logistic regression models binary outcomes using the log-odds, while linear regression predicts continuous values. Logistic regression outputs probabilities and uses a sigmoid function, making it suitable for classification tasks.