Корреляция и регрессия

Regression and correlation are related but distinct statistical techniques. Correlation measures the strength and direction of the linear relationship between two variables using the Pearson correlation coefficient, expressed as 'r'. Regression, however, establishes a predictive relationship by fitting a line to data points, allowing you to predict one variable's value based on another known variable.

The Pearson correlation coefficient 'r' ranges from −1 to +1. A value closer to −1 or +1 indicates a strong correlation between variables, while a value near zero suggests no linear correlation. Positive values indicate variables moving in the same direction, whereas negative values indicate inverse movement. The magnitude determines correlation strength regardless of sign.

Linear regression produces a best-fit equation that reliably predicts unknown variable values. In analytical chemistry, this equation can predict the signal value ('y') from a known measurement, or calculate the concentration of a substance ('x') from the signal it produces. This application is essential for calibration and quantitative analysis.

Correlation strength determines the reliability of a regression model's predictions. A strong correlation (r close to ±1) indicates the variables are tightly related, making the regression equation trustworthy for predictions. Weak correlation (r near zero) suggests the model may not reliably predict values, signaling the need to reconsider the relationship or data quality.

Positive correlation occurs when two variables move in the same direction—as one increases, the other increases. Negative correlation occurs when variables move in opposite directions—as one increases, the other decreases. Both can be strong (r values near ±1) or weak (r values near zero), depending on how closely the variables track their respective trends.

A correlation coefficient of zero indicates no linear correlation between two variables. The variables show no consistent directional relationship—changes in one variable do not correspond predictably to changes in the other. This does not mean the variables are unrelated; it only means their relationship is not linear.

In calibration, the correlation coefficient quantifies how well a linear relationship fits between known concentrations and measured signals. A high correlation coefficient (close to +1) confirms the calibration curve is reliable for converting future signal measurements into accurate concentration values, essential for quantitative analytical work.