Standard Pac and Linear regression

[Trond]

Hi everyone!
I am interested in the Standard Pac for HP 41CV, as I understand it includes the linear regression program from the standard applications manual. Is this pac any good otherwise? What else does it contain? (I don't have the standard applications book right now, although may be able to find it in storage somewhere) Does it contain anything not in the Standard Applications book?

[Joe Horn]

Standard Pac for which model calculator?

[Trond]

(11-11-2019 05:23 PM)Joe Horn Wrote:  Standard Pac for which model calculator?

Could have sworn I mentioned it, but I guess I didn't (OK I corrected the post above)
It's HP 41CV. I am thinking about getting another plug-in module.

[Valentin Albillo]

(11-11-2019 04:40 PM)Trond Wrote:  Hi everyone!
I am interested in the Standard Pac for HP 41CV, as I understand it includes the linear regression program from the standard applications manual. Is this pac any good otherwise? What else does it contain? (I don't have the standard applications book right now, although may be able to find it in storage somewhere) Does it contain anything not in the Standard Applications book?

In my opinion, it's pretty useless. The quality of the programs in it is mediocre to poor and there are much much better programs out there for linear regression.

For having some coding examples to see and study when you got your 41C for the first time it was barely adequate but otherwise I wouldn't waste a port, RAM, ROM or time to use it.

V.

[Trond]

(11-11-2019 07:31 PM)Valentin Albillo Wrote:  

(11-11-2019 04:40 PM)Trond Wrote:  Hi everyone!
I am interested in the Standard Pac for HP 41CV, as I understand it includes the linear regression program from the standard applications manual. Is this pac any good otherwise? What else does it contain? (I don't have the standard applications book right now, although may be able to find it in storage somewhere) Does it contain anything not in the Standard Applications book?

In my opinion, it's pretty useless. The quality of the programs in it is mediocre to poor and there are much much better programs out there for linear regression.

For having some coding examples to see and study when you got your 41C for the first time it was barely adequate but otherwise I wouldn't waste a port, RAM, ROM or time to use it.

V.

Good to know. I'm surprised, because the LR on my HP 34C is pretty good (from about the same time)

[Trond]

So...where can I find a good linear regression program for HP 41C?

[Ángel Martin]

(11-11-2019 11:28 PM)Trond Wrote:  So...where can I find a good linear regression program for HP 41C?

Take a look at the "SM44" entry in here: SandMath

[Valentin Albillo]

(11-11-2019 11:28 PM)Trond Wrote:  So...where can I find a good linear regression program for HP 41C?

It depends on what kind of "linear regression" are you interested in. For instance:

1. Just linear regression, as in y=a*x+b, or you'd also want power (y=a*x^b), logarithmic (y=a+b*ln(x)), etc ... fits ?
2. Just linear regression in one variable, as in y=a*x+b, or you'd also want linear regression for multiple variables: y=a0+a1*x1+a2*x2+a3*x3+...+an*xn ?
3. Just linear regression of degree 1, as in y=a*x+b, or polynomial regression of degree n: y=a0+a1*x+a2*x^2+...an*x^n ?
   

There are many possibilities with various degrees of complexity, you see.

V.

[SlideRule]

Any familiarity with CURVE FITTING FOR PROGRAMMABLE CALCULATORS {ISBN 8-943494-00-1} by William M. Kolb?

"PREFACE
This book provides all of the essential information needed to fit data to the most common curves. It avoids the usual mathematics and presents instead, straightforward solutions that can be used with most calculators. A basic introduction is included for the novice user of statistical models. The more intrepid explorer may find the sections on derivations, transformations, decomposition and substitution valuable for developing custom curve fitting routines. Users of the Hewlett-Packard HP-41C/V programmable pocket computer can make immediate use of many models presented with the program and bar code included …
…
1. REGRESSION
One technique used to formulate a statistical relationship among these variables is regression … The most common form of regression is …"

135 pages of documented & illustrated regression analysis & techniques applied to curve fitting on a programmable calculator.

BEST!
SlideRule

[Dave Britten]

(11-12-2019 11:59 AM)SlideRule Wrote:  Any familiarity with CURVE FITTING FOR PROGRAMMABLE CALCULATORS {ISBN 8-943494-00-1} by William M. Kolb?

"PREFACE
This book provides all of the essential information needed to fit data to the most common curves. It avoids the usual mathematics and presents instead, straightforward solutions that can be used with most calculators. A basic introduction is included for the novice user of statistical models. The more intrepid explorer may find the sections on derivations, transformations, decomposition and substitution valuable for developing custom curve fitting routines. Users of the Hewlett-Packard HP-41C/V programmable pocket computer can make immediate use of many models presented with the program and bar code included …
…
1. REGRESSION
One technique used to formulate a statistical relationship among these variables is regression … The most common form of regression is …"

135 pages of documented & illustrated regression analysis & techniques applied to curve fitting on a programmable calculator.

BEST!
SlideRule

That would probably be overkill, as it's a MASSIVE program that computes a few dozen fits simultaneously, but it's a fantastic reference if you want to use the formulas therein to implement smaller, more focused programs of your own.

[Trond]

(11-12-2019 06:09 AM)Valentin Albillo Wrote:  

(11-11-2019 11:28 PM)Trond Wrote:  So...where can I find a good linear regression program for HP 41C?

It depends on what kind of "linear regression" are you interested in. For instance:

1. Just linear regression, as in y=a*x+b, or you'd also want power (y=a*x^b), logarithmic (y=a+b*ln(x)), etc ... fits ?
2. Just linear regression in one variable, as in y=a*x+b, or you'd also want linear regression for multiple variables: y=a0+a1*x1+a2*x2+a3*x3+...+an*xn ?
3. Just linear regression of degree 1, as in y=a*x+b, or polynomial regression of degree n: y=a0+a1*x+a2*x^2+...an*x^n ?
   

There are many possibilities with various degrees of complexity, you see.

V.

I actually do have it for multiple variables. No, I need a quick number 1 in the list above (straight line, the others usually go by other names). I found the program from the Standard Pac, but then you said it's no good. I can do it on another HP of course (I'm really liking 34c), it's just I think I should have it on the 41CV as well, and this is the calculator I usually take to work.

[Dave Britten]

I use this simple program if I need plain old linear regression on my 41:

http://dave.brittens.org/HP41C/LinearRegression.txt

I can't remember if I wrote it or found it somewhere else and adapted it slightly. This works off the standard stat registers, so you don't have to use a special program to enter data, and it includes tiny routines for predicting x or y from the value of the other variable. It fits an equation of the form y=mx+b, and also computes correlation.

[KeithB]

Does it calculate the coefficient of correlation correctly? 8^)
http://www.talkorigins.org/faqs/c-decay.html

[SlideRule]

If your interest extends to the archives, then hp41 howto do simple linear regression? (Posted by David E on 17 Dec 2002) should have modest appeal.

If you use the sigma+ button (top left on the keypad)(after clearing the summation registers using the CLSigma button - shift x<>y key, just below the sigma+ key), you can accumulate the sums of x, x^2, y, y^2, and xy in memory registers 11 through 15 respectively, while register 16 will get the value of n (the number of x,y pairs). To put in your numbers, enter y first, then "ENTER", and then enter your x value - then hit the sigma+ key.
After you have entered all you pairs of values, you need to remember (or be told - or, left as an exercise for the student, derive!) the least-squares fitting expressions for slope, intercept, and correlation coefficient. In case you are rusty:

the slope is (n*SUMxy - SUMx*Sumy)/(n*SUMx^2 - (Sumx)^2)

the intercept has the numerator SUMy*SUMx^2 - SUMxy*SUMx (with the same denominator as for the slope)

r is (n*SUMXY - SUMx*SUMy) / ( SQRT(n*SUMx^2 - (SUMx)^2) * SQRT(n*SUMy^2 - (SUMy)^2) )

The above and more awaits your perusal.

BEST!
SlideRule

[Trond]

(11-13-2019 12:20 PM)Dave Britten Wrote:  I use this simple program if I need plain old linear regression on my 41:

http://dave.brittens.org/HP41C/LinearRegression.txt

I can't remember if I wrote it or found it somewhere else and adapted it slightly. This works off the standard stat registers, so you don't have to use a special program to enter data, and it includes tiny routines for predicting x or y from the value of the other variable. It fits an equation of the form y=mx+b, and also computes correlation.

Simple, and it works. Thanks!