Regression & Correlation Vocabulary

a measure of the strength and direction of relationship between variables

“Height and weight show a positive correlation.”

Origin: Medieval Latin correlatio, from com- `together` + relatio `relation`

a method for modeling the relationship between variables

“Linear regression predicts salary based on years of experience.”

Origin: Latin regressus `a going back`, from regredi `to go back`

the proportion of variance explained by a model (R²)

“An R² of 0.85 means the model explains 85% of variance.”

Origin: From coefficient + determination

the difference between observed and predicted values

“Analyzing residuals reveals patterns the model missed.”

Origin: Latin residuum `that which remains`, from residere `to remain behind`

non-constant variance in the error terms of a regression

“Heteroscedasticity violates key regression assumptions.”

Origin: Greek heteros `different` + skedasis `dispersion`

high correlation among predictor variables

“Multicollinearity inflates standard errors of coefficients.”

Origin: Latin multi- `many` + co- `together` + linearis `linear`

estimating values within the range of known data

“We used interpolation to estimate the value at x = 5.”

Origin: Latin interpolare `to refurbish, alter`, from inter `between` + polare `to polish`

estimating values beyond the range of known data

“Extrapolating too far from the data is risky.”

Origin: From extra `outside` + interpolation

an apparent relationship caused by a third variable or chance

“The correlation between ice cream sales and drowning is spurious.”

Origin: Latin spurius `illegitimate` + correlation

a relationship where one event directly produces another

“Correlation does not imply causation.”

Origin: Latin causatio `a causing`, from causa `cause`