All MCQs

Overfitting occurs when a model captures sample-specific noise in addition to genuine patterns. Its training performance is strong, but its generalization performance is weak.

Model fitting estimates parameters by optimizing a measure such as error or likelihood using observed data. The fitted model can then be evaluated or used for prediction.

The likelihood treats the observed data as fixed and compares how strongly parameter choices support those data. Model fitting often selects parameters that maximize this quantity.

Bayes’ theorem combines a prior probability with evidence represented by the test result. It produces the posterior probability of disease after observing the evidence.

Independence means P(A|B) equals P(A), assuming B has positive probability. Information about one event therefore provides no probability update for the other.

Conditional probability updates the probability of event A using the information that event B occurred. The vertical bar is read as “given.”

Standard deviation is the square root of variance, returning the measure to the variable’s original units. Variance is expressed in squared units, which are often less intuitive.

Variance quantifies dispersion by averaging squared distances from the mean. Squaring prevents positive and negative deviations from canceling.

Expected value weights each possible outcome by its probability and sums the results. Across many repetitions, the average outcome tends toward this value under suitable conditions.

The cumulative distribution function adds probability from the lower end of the distribution up to x. It therefore returns P(X ≤ x).

In a continuous distribution, probability is represented by area under the density curve over a range. The area associated with one exact point is zero.

A probability mass function assigns probabilities to individual values of a discrete random variable. Counts such as 0, 1, or 2 defects take distinct possible values.