Any non-vertical straight line (you'd be in a fine pickle if your weight trend line were vertical, wouldn't you?) can be expressed in the form:

*Y = mX + b*

where *m*
is the *slope*, giving the change in the *Y*
axis value
for each unit change along the *X*
axis, and *b*
is the *intercept*, the point at which the line crosses the *Y*
axis when *X*
is zero.

To find *m*
and *b* for the line that best fits a collection of data
points
*D*_{1}, *D*_{2}, ... *D*_{n}
we calculate:

Since we're only interested in the rate of change, we only need the
slope, *m*, which gives the daily rate of change in the line that best
fits the moving average trend curve. From the slope, the
average weight change per week is just seven times the daily change,

7*m*

and the average daily calorie deficit (if negative) or excess (if positive) is:

3500*m*

By John Walker