PHARM/PCS V5 · Procedure 3 Dosage and Statistics
Linear Regression I
Fit a least-squares straight line to paired x and y data and get the slope and intercept, or predict y at any x.
What it computes
Ordinary least-squares fit of y = mx + b to a set of paired (x, y) points.
Inputs and outputs
You provide
- Paired x and y values, one pair per line
- Optionally, an x value at which to predict y
PHARM/PCS reports
- Slope and intercept of the fitted line
- The correlation coefficient r
- The predicted y, if you asked for one
- A plot of the data with the fitted regression line
When to use it
When you want a quick straight-line fit and only need the slope and intercept, the streamlined version. If you also need the correlation coefficient, standard errors, and confidence limits, use Procedure 5, the full-diagnostic version of this same fit; if theory forces the line through zero, use Procedure 4.
Reading the results
Least squares assumes the scatter is in y (not x) and roughly constant across the range. Always eyeball the plotted points against the line, since a good-looking slope can still hide curvature or a single outlier that one r value won't reveal.
Want the full statistics?
This is the quick version of the fit. Procedure 5 runs the same regression and adds standard errors, the x-intercept, and confidence limits. The line matches the 1986 output to six significant figures.
This page describes the procedure. Calculations run inside the licensed PHARM/PCS V5 application. PHARM/PCS is a trademark of MCS Scientific LLC.