The Museum of HP Calculators

HP Forum Archive 21

[ Return to Index | Top of Index ]

Where is linear regression on an HP-41
Message #1 Posted by Palmer O. Hanson, Jr. on 12 Aug 2012, 10:50 a.m.

I wanted to do a linear regression on my HP-41CV. I did a Clear Sum and entered the data pairs using Sum+ . But I can't find the instruction to do a linear regression. I have looked in the manual and have SST'd through the catalog. So, what is the instruction for linear regression?

      
Re: Where is linear regression on an HP-41
Message #2 Posted by Dieter on 12 Aug 2012, 11:10 a.m.,
in response to message #1 by Palmer O. Hanson, Jr.

Palmer, there are no regression commands on the HP-41. That's why the the standard applications booklet that came with the '41 included such a program. For linear as well as non-linear regression. You will also find it here on hpmuseum.org in the HP-41 software section.

You just have to key it into your '41. Or get the standard pac, a ROM module that included all the programs in that booklet.

Dieter

            
Re: Where is linear regression on an HP-41
Message #3 Posted by Gerson W. Barbosa on 12 Aug 2012, 1:12 p.m.,
in response to message #2 by Dieter

Quote:
Or get the standard pac, a ROM module that included all the programs in that booklet.

The Advantage Pac is a better option. I don't know which one is more easily available though.

Gerson.

                  
Re: Where is linear regression on an HP-41
Message #4 Posted by Angel Martin on 12 Aug 2012, 6:18 p.m.,
in response to message #3 by Gerson W. Barbosa

The SandMath has MCODE functions for LR, Y, COV and CORR, all courtesy of Jean-Marc Baillard

forgot to mention: in this implementation I used 13-digit OS routines, no like the user code programs.

Edited: 12 Aug 2012, 6:29 p.m.

                        
Re: Where is linear regression on an HP-41
Message #5 Posted by Gerson W. Barbosa on 12 Aug 2012, 8:49 p.m.,
in response to message #4 by Angel Martin

Hello ┴ngel,

Quote:
forgot to mention: in this implementation I used 13-digit OS routines, no like the user code programs.

That's why I said 'a better option', not 'the best option' :-)

All the best,

Gerson.

      
Re: Where is linear regression on an HP-41
Message #6 Posted by Patrice on 12 Aug 2012, 2:41 p.m.,
in response to message #1 by Palmer O. Hanson, Jr.

Hi Palmer,

On a vanilla 41C, you only have the stat registers which gives you everything needed for a linear regression.

You have to apply the formulas by yourself, it is usually done with a little program.

It was the good time of self made tools :)

Patrice

      
Re: Where is linear regression on an HP-41
Message #7 Posted by Namir on 12 Aug 2012, 3:17 p.m.,
in response to message #1 by Palmer O. Hanson, Jr.

Did HP run out of ROM space to implement the linear regression calculations? The TI-59 (the rival of the 41C) had a special OP command that performed linear regression.

Namir

            
Re: Where is linear regression on an HP-41
Message #8 Posted by Dieter on 12 Aug 2012, 4:15 p.m.,
in response to message #7 by Namir

Yes, the TI58/59 used OPá12 ... OPá15 for linear regression, while the 41C did not provide such functions. On the other hand, do you think TI ran out of ROM space so that they did not include a factorial function, or something like e^x-1 and ln1+x, H.MS arithmetics, a remainder function, decimal-octal conversions etc. etc. ?-) All this is included in a standard 41C.

Calculating the two parameters of a linear regression is simple and straightforward. Providing an exact ln1+x function is not. So I'm glad HP used the available memory for the trickier cases. ;-)

Dieter

            
Re: Where is linear regression on an HP-41
Message #9 Posted by Mike Morrow on 12 Aug 2012, 7:25 p.m.,
in response to message #7 by Namir

Quote:
The TI-59 (the rival of the 41C)...

I doubt that many who were contemporary with the introduction of these machines considered the 1977 TI-59 to be a rival of the 1979 HP-41C. Most considered it to be the rival of the 1976 HP-67. If anything, the HP-41C was introduced as a rival to the TI-59.

IMHO, the HP-41C series was a next-generation trek into new territory and never had a comparable subsequent TI rival, in the sense of something that was created as a response to it and as widely distributed.

Edited: 12 Aug 2012, 7:26 p.m.

                  
Re: Where is linear regression on an HP-41
Message #10 Posted by Luiz C. Vieira (Brazil) on 12 Aug 2012, 8:34 p.m.,
in response to message #9 by Mike Morrow

Hi.

Allow me to add that I cannot see neither consider a reference to compare both. Apart of the fact they are both programmable scientific calculators, what I can think as differences in order to compare both are:

- arithmetic notation - AOS x RPN
- expandability, either peripheral devices or RAM/ROM
- system resources for programming (somehow related to previous item)
- display capability and technology
- battery life expectancy (somehow related to previous item)

Cannot think of any other major differences, considering the ones I pointed out are actually major. So, there are many different points setting them apart of each other, comparing both should not lead to fair conclusions.

Cheers.

Luzi (Brazil)

Edited: 12 Aug 2012, 8:35 p.m.

                  
Re: Where is linear regression on an HP-41
Message #11 Posted by Namir on 12 Aug 2012, 8:50 p.m.,
in response to message #9 by Mike Morrow

Mike,

I stand corrected. You are right! The TI-59 was the rival of the HP-67/97.

Namir

                  
Re: Where is linear regression on an HP-41
Message #12 Posted by Gerson W. Barbosa on 12 Aug 2012, 9:20 p.m.,
in response to message #9 by Mike Morrow

Quote:
IMHO, the HP-41C series was a next-generation trek into new territory and never had a comparable subsequent TI rival, in the sense of something that was created as a response to it and as widely distributed.

Might the never released TI-88 have been a valid competitor?

                        
Re: Where is linear regression on an HP-41
Message #13 Posted by Mike Morrow on 12 Aug 2012, 9:48 p.m.,
in response to message #12 by Gerson W. Barbosa

Quote:
Might the never released TI-88 have been a valid competitor?

That's exactly what I thought back then, and today still.

That's why I had to qualify my characterization of no TI rival to the HP-41C with "in the sense of something that was created as a response to it and as widely distributed." The TI-88 didn't meet those last three words. :-)

It's too bad...I was looking forward to the TI-88. But back then, I was even interested in the TI-99/4 and 99/4A.

                              
Re: Where is linear regression on an HP-41
Message #14 Posted by Namir on 13 Aug 2012, 12:03 a.m.,
in response to message #13 by Mike Morrow

The TI-95 is a very capable AOS machine. It takes one plug-in module and has a printer port. It can be seen as somewhat a competitor of the 41C, but not a serious rival.

Namir

                              
Re: Where is linear regression on an HP-41
Message #15 Posted by Gerson W. Barbosa on 13 Aug 2012, 1:28 a.m.,
in response to message #13 by Mike Morrow

I had a TI-59 for a while in '82, until it was stolen. I would surely have chosen the TI-88 as a replacement if it were available, since I was used to TI (previously I had a TI-51-III). I ended up getting an HP-15C instead because of its looks and because it was less expensive. At first I was disappointed because there were no equals keys. Only at first, though :-)

                  
Re: Where is linear regression on an HP-41
Message #16 Posted by Palmer O. Hanson, Jr. on 12 Aug 2012, 10:58 p.m.,
in response to message #9 by Mike Morrow

Quote:
I doubt that many who were contemporary with the introduction of these machines considered the 1977 TI-59 to be a rival of the 1979 HP-41C. Most considered it to be the rival of the 1976 HP-67. If anything, the HP-41C was introduced as a rival to the TI-59.

IMHO, the HP-41C series was a next-generation trek into new territory and never had a comparable subsequent TI rival, in the sense of something that was created as a response to it and as widely distributed.


I was heavily involved in the so-called "friendly competition" back in those days. The way I remember it is

The HP-67 and the TI SR-52 were roughly comparable in calculating capability. The major advantage of the SR-52 was the ability to connect to the PC-100 printer.

The TI-59 clearly outclassed the HP-67 with three major advantages: more user memory, the solid state modules and the ability to connect to the PC-100 printer.

The HP-41 clearly outclassed the TI-59 with major advantages such as LCD rather than LED display, speed, and more user memory, particularly with the use of memory modules with the HP-41C and more internal memory with the HP-41CV and HP-41CX.

The TI-88 was expected to be a competitor to the HP-41 series but never made it to market. There were some problems with the hardware, and in particular with the keyboard which was of the same family as the TI-55-II. At the same time TI abandoned the TI-59 in favor of emphasis on trying to dominate the early PC market with the TI-99/4.

Eventually, TI tried to re-enter the high end scientific calculator market with the TI-95. It's major disadvantage was it's size. Two major advantages relative to the HP-41 were speed and the programmable memory modules. After only a few years the TI-95 was abandoned when TI decided to enter the education market with the TI-81. It would be difficlt to fault them for that decision from a corporate standpoint.

Back to the availability of linear regression. I don't have most of my collection here but I thought that I remembered that some of the earlier machines did offer linear regression as a keyboard option. I reviewed the descriptions for a few of the machines using the descriptions in this Museum and found that the HP-22, HP-27, HP-32E and HP-33E did have linear regression as a keyboard options and the HP-25, HP-29 and HP-67 did not. My personal preference is for a keyboard capability.

                        
Re: Where is linear regression on an HP-41
Message #17 Posted by Mike Morrow on 13 Aug 2012, 2:15 p.m.,
in response to message #16 by Palmer O. Hanson, Jr.

Quote:
The HP-67 and the TI SR-52 were roughly comparable in calculating capability. The major advantage of the SR-52 was the ability to connect to the PC-100 printer.

I always considered at the time the 1975 SR-52 to be the rival of the 1974 HP-65. Other than the PC-100 capability of the SR-52, it had nothing to recommend it over the 1976 HP-67...and never would I consider it otherwise comparable to the HP-67.

Quote:
The TI-59 clearly outclassed the HP-67 with three major advantages: more user memory, the solid state modules and the ability to connect to the PC-100 printer.

The 1977 TI-59 was very impressive technically, compared to the HP-67. But its reliability was terrible...I've recounted here many times before that I used up five TI-59s and three PC-100s between 1977 and 1981, during which time I only had to send in my HP-67 once, in 1980. When I had to have some hardware that I knew would work such as for submarine propulsion and navigation applications, I had to have the HP-67 or 97.

The industrial design and color scheme of the SR-50A through the TI-59 was also terrible aesthetically. I wonder if that is what "inspired" HP to adopt similarly ugly poor-contrast color schemes for the HP-48, -38G, and Pioneer series.

                              
Re: Where is linear regression on an HP-41
Message #18 Posted by Palmer O. Hanson, Jr. on 13 Aug 2012, 11:28 p.m.,
in response to message #17 by Mike Morrow

Quote:
I always considered at the time the 1975 SR-52 to be the rival of the 1974 HP-65. Other than the PC-100 capability of the SR-52, it had nothing to recommend it over the 1976 HP-67...and never would I consider it otherwise comparable to the HP-67.
Curiously enough, when the HP-67 was released the editor of 52 Notes wrote much the same sort of thing about the HP-67 -- disappointing, and no reason to buy it.
Quote:
The 1977 TI-59 was very impressive technically, compared to the HP-67. But its reliability was terrible...I've recounted here many times before that I used up five TI-59s and three PC-100s between 1977 and 1981, during which time I only had to send in my HP-67 once, in 1980. When I had to have some hardware that I knew would work such as for submarine propulsion and navigation applications, I had to have the HP-67 or 97.
I have never understood all the horror stories about the TI-59. I worked at Honeywell. The company participated in TI's Productivity program where any employee could get a TI-59 of his own by taking a short introductory course. So there were literally a thousand or so TI-59's around. The one problem which could give users the fits was the rarely occurring memory problem. I wrote a more thorough version of the memory test in the Service manual to check for that.

I had one friend who refused to take the course and get his TI-59. He insisted that his ditsy little HP-25 was better because it used RPN. We all thought that was more than a little odd.

During the period from the release of the TI-59 to the release of the HP-41 I knew of no one who thought that the HP-67 was a serious competitor for the TI-59. The editor of 52 Notes had a comment on that. He said that the HP users looked at their machines and thought "It just has to be better since I paid so much more."

The HP-41 was clearly better than the TI-59 in many ways. But I didn't like RPN and I worried about the accuracy of ten digit calculations instead of 13. Furthermore, I had purchased a Radio Shack Model 100 which could support some serious plotting capability and had switched to it for most of my work in inertial navigation until I retired. I didn't get an HP-41 until I started collecting. Now I have four HP-41's: a C, a CV, a CX and a C with the gold balls and two of the calculating errors. I also have both a HP82240A and HP82240B printer. I have to admit that all that hardware doesn't get very much use.

                                    
Re: Where is linear regression on an HP-41
Message #19 Posted by Mike Morrow on 14 Aug 2012, 12:49 p.m.,
in response to message #18 by Palmer O. Hanson, Jr.

I wrote: "Other than the PC-100 capability of the SR-52, it had nothing to recommend it over the 1976 HP-67...and never would I consider it otherwise comparable to the HP-67."

Quote:
Curiously enough, when the HP-67 was released the editor of 52 Notes wrote much the same sort of thing about the HP-67 -- disappointing, and no reason to buy it.

That editor's statement was obviously unburdened by the weight of facts. It would be true in a comparison of the HP-67 to the TI-59...but not to the grossly less capable SR-52.

I had my ship purchase a SR-52/PC-100A for use in the nucleonics lab in early 1976, before either the HP-67 or TI-59 was available. When the HP-67 appeared later, I found it to be a great advance over the SR-52 in every way except printer capability and (as you mention) the greater accuracy of the SR-52. At the cost of more than six weeks of my lieutenant(jg) pay I bought a personal HP-67, my first HP, with a few accessories. The SR-52 was clearly out of the contest, even though my personal calculators before then were all TIs purchased from 1974 to 1976...the SR-50, SR-51A, and SR-56.

The TI-59 in 1977 was a stunning product at a very competitive price (except for its ugly industrial design and color scheme). If it had been available before I purchased the HP-67, I doubt that I would have purchased an HP-67 until experiencing all those many failures of TI quality. I bought my personal TI-59/PC-100A in 1977, and I liked it greatly because of its capabilities. The TI-59 $300 price in 1977 was equivalent to almost $1200 today, compared to the HP-67's $450 price, equivalent today to $1700.

Quote:
I have never understood all the horror stories about the TI-59.

The only thing to understand is that quality of manufacture was very obviously lacking...failures occurred under very benign and non-challenging environmental conditions. My first TI-59 died within a year. The replacement died within six months, as did the subsequent replacement and my first PC-100A. When in 1981 my fourth TI-59 and second PC-100A died, I wrote the Chairman of TI, Mark Shepherd, describing my personal history with this product. I'm sure he never saw the letter, but I received a letter from an underling stating that a newly-made TI-59 and PC-100 would be tested and burned-in and then sent to me without cost (I had to pay for some of the earlier replacements), along with the address of a TI engineer who would be examining my failed units. The printer replacement was the less-capable PC-100C.

I required hardware to serve reliably for fairly significant submarine reactor, nucleonics, equipment performance monitoring, and navigation applications. The TI-59 failed repeatedly...the HP-67 did not. That was a real shame, because the TI-59/PC-100A capabilities were better suited to my needs than those of the HP-67, even after taking into account the inefficiency of the TI AOS which resulted in programs typically at least 40 percent longer than when implemented on the HP-67.

Quote:
I worked at Honeywell. The company participated in TI's Productivity program where any employee could get a TI-59 of his own by taking a short introductory course.

I read of those programs with envy back then. TI had no such offerings for US Navy people. :-)

Quote:
I had one friend who refused to take the course and get his TI-59. He insisted that his ditsy little HP-25 was better because it used RPN. We all thought that was more than a little odd.

Indeed. The HP-25 was useful only for a very very limited set of applications, compared to those possible on the TI-59. And TI never was so incompetent as HP to design a battery charging system that was guaranteed to destroy the calculator under conditions that were bound to occur naturally and inevitably...a shameful failure of HP to meet a trivial engineering problem from the here-so-revered mythological Bill Hewlett era!

Quote:
During the period from the release of the TI-59 to the release of the HP-41 I knew of no one who thought that the HP-67 was a serious competitor for the TI-59. The editor of 52 Notes had a comment on that. He said that the HP users looked at their machines and thought "It just has to be better since I paid so much more."

Experience...the best teacher...taught me how wrong sophomoric conclusions as these from that editor can be. In fairness, it is not likely that many TI-59 users had applications that required use in real-time to reliably produce results that then were utilized immediately in a major process, as my applications (like a reactor high power calorimetric, or a sea-traffic contact motion analysis) often required.

I spent a lot more "hobby time" with my TI than with my HP. It was a much more interesting system when it worked. I really liked the TI-59, but I never lost sight of its oft demonstrated lack of reliability. I stopped writing programs for the TI that were required for professional purposes.

Quote:
The HP-41 was clearly better than the TI-59 in many ways. But I didn't like RPN and I worried about the accuracy of ten digit calculations instead of 13.

I'd say that the HP-41C series was better than the TI-59 in every way except the accuracy issue. That issue remained with all HP machines until the appearance of the Saturn-based models...so HP recognized it eventually.

Quote:
I had purchased a Radio Shack Model 100 which could support some serious plotting capability and had switched to it for most of my work in inertial navigation until I retired.

The Model 100...that's a real milestone from 1983 regardless of its slow speed and the old-even-then 80C85 uP. I've got a couple of them that I used professionally in commercial nuclear plant engineering applications between 1984 and 1987. I loved its double-precision Basic in which some surprisingly sophisticated applications could be implemented.

Edited: 14 Aug 2012, 1:42 p.m.

                                          
Re: Where is linear regression on an HP-41
Message #20 Posted by Garth Wilson on 14 Aug 2012, 6:05 p.m.,
in response to message #19 by Mike Morrow

I wonder if TI improved the quality a lot after the initial years. I got a TI-58c in Dec '81 and later a 59 and used them both heavily, also the 100c printer, and never had any trouble with them except that after many years the 59's rubber wheel in the card drive deteriorated as they all do. In 2003-2005 our son used my 59 in high school. (I had sold the 58c before that.)

As for appearance, I initially thought the HP-41 was kind of funny-looking (with its flat end where the modules plug in) compared to the TI-59. I guess beauty is partly in the eye of the beholder though. To me, styling peaked about 1980 or soon after, and has been going downhill ever since, whether it's calculators, cars, appliances, stereos, whatever. It's ok that a friend thinks his wife is beautiful too. I won't tell him how wrong he is. LOL

      
Re: Where is linear regression on an HP-41
Message #21 Posted by Namir on 13 Aug 2012, 7:51 a.m.,
in response to message #1 by Palmer O. Hanson, Jr.

Of course the STAT 1 module made up for the lack of built-in linear regression by providing for linear regression for 2, 3, and 4 variables as well as quadratic and cubic polynomial regression.

Namir

            
Re: Where is linear regression on an HP-41
Message #22 Posted by Palmer O. Hanson, Jr. on 13 Aug 2012, 10:18 p.m.,
in response to message #21 by Namir

Quote:
Of course the STAT 1 module made up for the lack of built-in linear regression by providing for linear regression for 2, 3, and 4 variables as well as quadratic and cubic polynomial regression.
My records show that I have the following modules:

MathPac with manual

14001 MathPac with manual

15002 StatPac with manual

Financial Decisions Pac with manual

Thermal Pac without manual

Unfortunately I only have the MathPac with me and I didn't bring the manual.

Palmer

                  
Re: Where is linear regression on an HP-41
Message #23 Posted by Gerson W. Barbosa on 13 Aug 2012, 10:43 p.m.,
in response to message #22 by Palmer O. Hanson, Jr.

Quote:
Unfortunately I only have the MathPac with me and I didn't bring the manual.

The manual is available as TOS.

Table of Contents:

Matrix Operations Solution to f(x) = 0 on an Interval Polynomial Solutions/Evaluation Numerical Integration Differential Equations Fourier Series Complex Operations Hyperbolics Triangle Solutions Coordinate Transformations

Gerson.


[ Return to Index | Top of Index ]

Go back to the main exhibit hall