On 2002-03-27 12:13, Aladin Akyurek wrote:

* Hi Ian, Aladin, and Joe:*

Thanks Aladin and Joe for scholarly input.

How about use of Numerical Methods, .. and

how about looking at creating a formula from the Celsiuns and Farenheit data, say for lack of better words, by reverse engineering.

By looking at enough data (and with some assumptions of course), one may recognize that it will fit the straight line equation formula ...

y=mx+c ... with x as the slope and c as the offset

Could one not then come up with the best fit for value of x and c that will fit the data.

I don't know if it is relevant here ... but I thought I will mention it any way!
I take the gist of Ian's question to be to return a formula after examining a set of numbers by means of formulas and/or VBA code. Considering your example, we should have a system of formulas and/or WBA code that return an Excel formula, equivalent of y=mx+c.

Here a set of heuristics (citing from Langley et al.)

(1) If the values of a term are constant, then infer that the term always has that value.

(2) If the values of two numerical terms increase together, then consider their ratio.

(3) If the values of one term increase as those of another decrease, then consider their product.

BACON.1 has the above heuristic rules in the form of

*productions* as they are called in artificial intelligence. BACON, given a set of emiprically obtained numbers relating to the distances between, say, 3 planets, attempts to discover Kepler's third law of planetary motion: that is,

D^3/P^2 = c, where D is the distance, P is the period, and c is a constant.

Aladin

## Like this thread? Share it with others