Linear regression · Downward trend
The slope retains its sign and points remain original samples.
Compare samples with an ordinary least-squares fit y = slope × x + intercept. A linear relationship is a modelling assumption, not a forecast or causal conclusion. No extrapolation is drawn.
Up to 8 series and 4096 total XY rows. IDs are stable and at most96 characters; ols: is reserved. Incomplete pairs remain unknown and are excluded from fitting. Two distinct known X values are required. Constant Y gives slope zero and R² undefined. Calculations retain original input units.
Original samples and derived ols:<seriesId> fits share one axis frame with independent legend visibility and colours. Numeric XY zoom clips the original fitted line without retraining it. Point details identify estimates, coefficients, known/unknown pair counts and R². Shared tooltips, stacking and fitted confidence intervals are unavailable.