HP Prime - Statistics

[paul.hermans@benemtech.com]

Hi all,
I am trying to prepare for a test, and the study guide has a problem that is as follows:

What is the area between z=-1 and z=1

I know that the answer is 0.683, but wondering how I would use the Prime to calculate this for any interval.

the normald_cdf( mean, mode, x) only gives me the left tail up to the value of x.....so i could do it twice and subtract, but that seems crazy....is there a better way?


Thanks in advance

Paul

[walter b]

(08-22-2014 04:00 AM)paul.hermans@benemtech.com Wrote:  I am trying to prepare for a test, and the study guide has a problem that is as follows:

What is the area between z=-1 and z=1

!

d:-/

[walter b]

Paul,

I've read your OP and it was even 'nicer'. IF you really want to pass a test in statistics, I'd recommend learning statistics and a bit of math well in advance instead of learning to press certain keys on whatever calculator you were trained on.

You didn't even specify the function you wanted to know the integral of (but it was easy to guess from the context). The normal distribution function is symmetric - that allows calculating a solution to your problem easily.

Good luck to you and your company!

d:-/

[paul.hermans@benemtech.com]

To clarify for Walter whose answer was about as useless as any i have ever seen as it did not answer the question. I will rephrase the question to better explain what i need help with.

Background:
I have a new calculator that I am trying to learn to use. It is an HP Prime. The manual sucks and makes it hard to find anything of use, and on calculator documentation is a joke. In the manual, if i don't already know the name of a function, good luck finding it.

Current Situation:
I have to use this calculator on a test of which statistics is a very small portion of the material covered. The test has limited time, and I am trying to figure out how to use the Prime to answer certain questions QUICKLY.

I forgot to mention that I was referring to a Normal Distribution but my understanding was that Z-Scores are also sometimes referred to as standard scores and are always part of a normal distribution (not sure this is correct, but that is why i didn't spell out that it was a normal distribution)

So, how do i calculate the area under the curve (yes, the normal distribution curve) between two values of Z. Yes, if i had a definition of the function f(x) = ?? then I could try to integrate, but my memory tells me that was a very time consuming and difficult option and leads to nothing useful for this task. Yes, I could look up values in a table, but no table will be provided. Yes, I could memorize certain values like .683 being between +- std. deviation and estimate, but that is pointless. So what am i left with? Oh, using the calculator to calculate it.....silly me, why didn't I think of that?

There is a function in the HPPrime called normald_cdf(mean, std dev, value) which calculates the area under the curve to the left of the value specified. So, back to my original question, is there a faster/better way to calculate the area between two values for this distribution?

I could do normald_cdf(mean, std dev, upper_value) - normald_cdf(mean, std dev, lower_value)
but that takes forever to enter, so without asking for more condescending, useless answers, is there a way to use the HP Prime to do this faster?

For reference, on the TI calculators a function exists that takes an additional argument so i can specify the upper and lower bound of the region i want to calculate the area for, i am looking for something like this in the Prime.

Thanks in advance.


PS - this test has nothing to do with my company, but thanks for assuming.

[Tim Wessman]

You are correct - Prime does not currently support four arguments to do a "between" calculations on the probability distributions. That is on the list of potential future changes however.

As to the manual, are you looking at the full manual? Or just the quick start guide? Please give detailed suggestions as to how to resolve "that it sucks". The full manual seems quite well organized to me. So understanding where you are looking or HOW you are looking would be quite helpful.

You also mention the on calculator help sucks, yet it is definitely the most comprehensive on-system help of ANY calculator out there. For this question, I press HELP, then the TREE, then SEARCH. Type NORM and then with a few clicks have found the normal distribution functions. Inside them, a single tap takes you to the related functions. So again, how were you trying to use it that you feel it is a joke? How or what should be changed to make that better? Describing your thinking in trying to find what you are looking for would be helpful.

Now if you really are concerned about making some of these calculations "faster", the immediate solution that comes to mind is to use the DEFINE function to make a user function to do it. Give it a name like NORM4, put in your function subtracting the two, and you now have it.

Another option would be to save a copy of the solve application, save the same equation, and then you can just go in there to solve it.

[David Hayden]

(08-22-2014 01:19 PM)paul.hermans@benemtech.com Wrote:  I could do normald_cdf(mean, std dev, upper_value) - normald_cdf(mean, std dev, lower_value) but that takes forever to enter.

Note that the first 2 arguments are optional and default to 0 and 1 respectively, so the formula you're looking for is normald_cdf(upper_value) - normald_cdf(lower_value).

You can enter the function by navigating from the toolbox key. I don't have my Prime with me right now, but I think after pressing the toolbox, it's "Probability" then "Cumulative" then "Normal".

Another possibility is to assign "normald_cdf" to a key and use the user keyboard.

Dave

[paul.hermans@benemtech.com]

Thanks Tim,
That does help, and your criticisms of my post are fair too....guess that is what happens with frustration.

It is not fair to say that the help document sucks, it is quite comprehensive in terms of describing the various functions. I think my frustration came from it being hard to find anything quickly. This really isn't because the manual is not good, but that the calculator does so much that it is quite long. For this example, since I didn't know the actual name of the function in the HP calculator, I searched for "Normal" which of course came up quite often. Where I got bogged down is the Inference App which does something with confidence intervals (which I haven't learned about yet so it is like a foreign language) but the word Normal appears a lot. Anyways, I did find it and it worked as documented.

I commented on the on calculator help in error as well. Silly as it seems now, I missed that when I hit help and then TREE, there was a search option.....i was scrolling through lots of commands. Dumb on my part.

One area where I think a bit of improvement could happen is the "explaining" how the different pieces of the calculator work together. I have a couple of examples:

1) If I have input an equation, say something like x^3-4x^2-5x+4 =0 and i want to evaluate that for x=5 and x=-2, how do I do that? I am 100% sure there is a way, but it is elusive so far. I would hope that it was easy to find in the manual but not yet.

2) I was using the CAS to do some basic stuff like factoring a polynomial....which worked great I might add, and then I decided I wanted to plot the function. Now I have trouble again. I selected the equation from the command line, and then selected the soft key for Copy.....pressed APP, selected Advanced Grapher and attempted to paste the equation in.....there is no soft key for Paste...so tried SHFT-MENU and it shows a list to paste from, but not the thing i just copied. Now i realize that the soft key for COPY only copied from the stack to the active input line....so i go back and use SHFT-VIEW to copy it, then back to the Adv Grapher and paste it in for V1. Click Enter - SYNTAX ERROR. So I try entering a Y= in front of what I pasted.....SYNTAX ERROR.....so I manually enter the equation from scratch....No problem...works like a champ. Turns out that for some reason an X in the CAS is not the same as an X in the advanced grapher....no idea why, no explanation about how this is supposed to work (at least not that I've found yet)

When I lashed out at the manual, problem 2 in particular was a source of that feeling. Why would y=x^2 from the CAS be any different than y=X^2 from the soft keys in the app? If there is a logical reason (there usually is) then that is what I would want to understand from a manual....how the calc works....how i move between various different things like CAS and an APP and if there are any limitations to doing so that i need to be aware of.

Does that make sense? A lot of the frustration is because I haven't used a calculator in a long time, the last one I used was an HP 28C and it was the new best thing ever back then.....and things have changed a little.

Thanks in advance for any guidance you can give on this type of thing.

[Bill Zimmerly]

I'm glad that Paul's frustration eventually led to Tim's reply ... I learned -A LOT- by the answers. Thanks Guys!

(Example: I never noticed the "Search" functionality under the "Help" tree before.)

- Bill

[Angus]

Paul, it seems to me that you miss a detailed hands on training for different usage and problems in Science, Engineering and mathematics. I do, too and others aswell.
The Problem with it could be that there is no such Template available for hp. The manuals can be modified for different calcs, but the manuals do not contain helpful explanations, Cross refrences and inspiration.
Providing a manual just as you would like to hold in your hands would be time consuming what means expensive. I think in terms of selling numbers the managment is not willing to do such an Investment.

[CR Haeger]

Hi Paul

There are a lot of helpful guides in the hp prime section in a sticky post.

I agree with you that it is not very intuitive to easily pass objects around (like expressions or equations) between the applications, Home and CAS screens. This is where I usually run into trouble.

Best,
Carl

[parisse]

normald_cdf accepts 4 arguments in Xcas, it should therefore work from CAS, unless I did the modification after the last firmware (I don't remember).

[paul.hermans@benemtech.com]

parisse,
I'm afraid I don't understand what you're telling me....is the help document wrong about how the function works? I tried with 4 arguments and get a syntax error.
what is Xcas? How is it different from CAS? Thanks for any info you can provide.

Did you personally change the way the command works? If so, how does one go about doing that?

Thanks

[parisse]

The Prime CAS has the same CAS kernel as Xcas. Xcas is updated much more frequently and a few commands of Xcas are not enabled inside the Prime. This could explain why you get a syntax error with your current firmware on a command like normald_cdf(1,2,0.5,2.5)

[aurelio]

(08-23-2014 05:27 AM)Bill Zimmerly Wrote:  I'm glad that Paul's frustration eventually led to Tim's reply ... I learned -A LOT- by the answers. Thanks Guys!

(Example: I never noticed the "Search" functionality under the "Help" tree before.)

- Bill

that's the greatness of this forum there's always a chance to learn (at least for me) something more.

[StephanP]

Maybe it's something new in the latest OS, but I just use NORMALD_CDF with four parameters:
NORMALD_CDF(mean, std dev, lower value, upper_value)

E.g. NORMALD_CDF(78, 0.76, 79, 1E99)
And it returns a nice probability value 0.0941...