View Full Version : The Table Function on the TI-30X MultiView is Extremely Useful

colemstpeter

09-30-2009, 10:08 AM

If anyone isn't using the TI-30x MultiView calculator for this exam I would highly recommend trying it out. At first, I was hesitant to abandon the familiar layout and feel of the TI-30XII but now I fully believe in the MultiView.

I think the "meatiest" feature is the Table Function. Basically, you can just enter any y=f(x) function and look at a table of x and y values. I was using it this morning on ASM 53 to quickly compute the CDFs of discrete distributions used in the inversion method for simulation.

I have used it in the past to find the roots of nasty equations to get answers when I lacked the insight to see the elegant and simple two line solution.

Rant done, thanks.

daaaave

09-30-2009, 10:15 AM

The table function is also useful when computing F*(x) for a Kolmogorov-Smirnov test. The more that you play with a multiview, the clearer it becomes that it is the best calculator for 4/C.

no driver

09-30-2009, 10:22 AM

Could you take a moment and write out a specific example of this usage for the rest of us?

I've been using the Stat functions for empirical stuff. Basically any time I need to know the variance of a data set or even the mean if there are more than a few values.

daaaave

09-30-2009, 10:39 AM

Ok, say we are looking at SOA problem # 172, where we want to do a K-S test with the null hypothesis that F(x)=1-\frac{1}{(1+x)^4} and our data values are 0.2, 0.7, 0.9, 1.1, and 1.3. On the Multiview, you can push the Table button, it asks for y as a function of x, so you type in y=F(x), and then the next thing that comes up is it asks for the increments. Set that to manual, and then on the table you just type in the 5 x values and it tells you the 5 values of F(x). There is an old exam problem with 10 data values, so you save even more time there. You can do the same thing with the inversion method.

The stat functions are also huge, and you should know how to find the unbiased estimate of the variance of weighted data on your calculator, but that can be done on the 30XIIS as well as the Multiview.

For those of you in my online seminar, I will try to put together a calculator tips video at some point late next week.

no driver

09-30-2009, 10:55 AM

Ok, say we are looking at SOA problem # 172, where we want to do a K-S test with the null hypothesis that F(x)=1-\frac{1}{(1+x)^4} and our data values are 0.2, 0.7, 0.9, 1.1, and 1.3. On the Multiview, you can push the Table button, it asks for y as a function of x, so you type in y=F(x), and then the next thing that comes up is it asks for the increments. Set that to manual, and then on the table you just type in the 5 x values and it tells you the 5 values of F(x). There is an old exam problem with 10 data values, so you save even more time there. You can do the same thing with the inversion method.

The stat functions are also huge, and you should know how to find the unbiased estimate of the variance of weighted data on your calculator, but that can be done on the 30XIIS as well as the Multiview.

For those of you in my online seminar, I will try to put together a calculator tips video at some point next week. I'm not quite sure how that will work as the people at TI who gave James the BA-II+ emulator haven't helped us get one for the Multiview, so I might have to use my ancient digital camera to record a video somehow.

To try this example I put in the function, chose Start=0, Step=1 and Ask-x. It only let me put in the first three values. Is there a way to keep adding more values or do you have to keep swapping out until you've seen all of the results?

colemstpeter

09-30-2009, 10:58 AM

Try Start = .1, Step = .1, Auto. Then you can scroll down and get each value, but you will get values you don't need.

Coquelicot

09-30-2009, 02:15 PM

For those of you in my online seminar, I will try to put together a calculator tips video at some point late next week.

SWEET! I can't wait...

uclatommy

09-30-2009, 03:32 PM

I'm going stop by Staples after work so I can pick up this calculator. Thanks for the tip. I waste lots of time repeatedly calculating the same function when tabulating stuff.

no driver

10-01-2009, 11:18 AM

I'm going to put all my TI-30XS calculator tricks in this thread. If you see anything that could be done better or you've got a different method of approaching a problem please share. Eventually I'll be glad to summarize all of these tricks (as well as those submitted by other posters) and add them to the sticky.

Instead of entering a long equation that might include typos, I use the [data] lists to calculate and sum my Chi-square test statistic.

1) [data]->L1 enter your expected occurrences

2) L2 enter your observed occurrences

3) With your cursor in L3, [data]->FORMULA->Add/Edit Frmla

4) Enter: ([data]->L2 - [data]->L1)^2/[data]->L1. When you are done it should look like L3=(L2-L1)^2/L1

5) [2nd][stat]->1-Var Stats->DATA=L3 FRQ=ONE->CALC

6) Find ∑x and compare it to the Chi-square table

If the problem only has a few groups of data it might be faster to do it by hand but on a problem with several groups of data you'd have to break the sum into multiple chunks to enter it into the calculator anyway and there's more chance to make a mistake with missed parentheses etc.

With this method if you notice a typo in your entered data in L1 or L2 you can correct that entry and it will recalculate the corresponding value of L3 automatically. You'll still need to recalc the stats to get the new sum of L3.

colemstpeter

10-01-2009, 04:37 PM

I'm glad that no driver brought up the missed parentheses problem, the MultiView has an amazing display that prevents these types of errors. It has a MathType font that is similar to expensive graphing calculators. You can enter the formula exactly as you would write it on paper. Fractions, exponents, parentheses, ect... are all clearly displayed and put in their appropriate positions.

Furthermore, while in the middle of an expression you can scroll up into the history of calculations and bring down previous expressions into your new one and then continue with anything else you needed.

Like I said before, if you haven't seen this calculator yet you have to check it out.

Also, I am all for people posting calculator tips. Thanks!

uclatommy

10-01-2009, 04:52 PM

People who use this calculator have a clear advantage to those who don't. It's almost unfair. The only gripe I have with it is that it looks like it was designed by fisher price. It is otherwise, bar none, the best calculator for this exam.

badmaj5

10-01-2009, 05:50 PM

I too am a big fan of this calculator. My one gripe is that the scrolling is much slower than on the previous calculator. For example if you are typing a long equation and find out that you entered a + instead of a - at the very beginning, it takes a long time to scroll all the way back to the beginning of the typed in line.

actuarialmath

10-02-2009, 10:42 AM

Is this calculator this one?

http://www.actuarialbookstore.com/product_details.asp?prod_id=453061233

badmaj5

10-02-2009, 10:43 AM

Yes.

no driver

10-02-2009, 10:46 AM

Is this calculator this one?

http://www.actuarialbookstore.com/product_details.asp?prod_id=453061233

Yes. However, you can probably find it cheaper than $20 in a local Office Max/Wal-Mart/etc and you won't have to wait for it to arrive in the mail.

actuarialmath

10-02-2009, 10:59 AM

Yes. However, you can probably find it cheaper than $20 in a local Office Max/Wal-Mart/etc and you won't have to wait for it to arrive in the mail.

Good point. I don't want to wait for too long. I will get it from local store.

OldOne

10-02-2009, 01:11 PM

Has anyone else had the problem with the Multiview where it doesn't accept all your keystrokes? Many times in the past month I've gotten to the end of a long problem, tried to figure out why I've gotten the wrong answer, & found when I scoll back up that one or two decimal places are missing in a long formula. I know I hit those keys, & I've never had this problem with the IIS. I'm wondering if I have a defective calculator, or if this is just a problem with the Multiview. If so, I may go back to the IIS - there's no time on the exam for issues like that.

Honestly I have had the same issue with this calculator last sitting so I ended up not using it, because that is not something I have time for during the exam. But with these new tips I may end up using it for the shortcuts and the other one mainly...just make sure I am pressing the numbers hard enough to get entered...

daaaave

10-02-2009, 01:50 PM

I initially had problems with my BA-II+ Pro not acknowledging all of my keystrokes, but eventually just learned to push buttons more firmly and the problem went away. I haven't had a similar problem yet with the MultiView.

no driver

10-02-2009, 01:53 PM

Mine periodically freezes and redraws the screen when I'm putting in equations. During the freeze it doesn't record anything I hit, so I wait for it to finish and then continue. Fortunately it seems to happen less frequently when the memory is cleared, which will be true on exam day.

On the off chance that it was due to mathprint mode, I switched to classic for a while (which also clears at least the scroll back buffer in the memory), but after using the calculator for a while in classic mode I started having the same problem again, so I'm using mathprint mode once more.

I had asked about this before and no one else seemed to have the same problem so I wasn't sure if mine was defective or what. Since other folks are having it too maybe TI should have used a more powerful processor or something on this model... Or maybe there are some that are defective.

OldOne

10-02-2009, 01:54 PM

I haven't figured out yet if it's that I'm not pressing the keys hard enough, or if it's more of a memory problem - I'm leaning towards that it's a memory thing when I'm inputting quickly.

OldOne

10-02-2009, 02:24 PM

Just talked to tech support at TI - they are adamant that it's user error & there is plenty of memory on the unit. Don't know what I was expecting...

Clam Chowdah

10-02-2009, 11:19 PM

Got this yesterday.

I too am a big fan of this calculator. My one gripe is that the scrolling is much slower than on the previous calculator. For example if you are typing a long equation and find out that you entered a + instead of a - at the very beginning, it takes a long time to scroll all the way back to the beginning of the typed in line.

You can just hit [2ND] [Left] or [Right] but I'll agree that using this calculator is a slower experience than using the TI-30II but that's probably just because I have to unlearn some of the "bad" habits I've picked up over the years as I've raced though problems. One drawback that I think is legitimate is the variable storage method of punching the key at most seven times in order to recall your stored numbers. You use ffewer keystrokes when compared to the TI-30II(or whatever it's called) and during the retrieval process the user can view the values right beneath the (A B C D E) labels. Contrast to the TI-30XS where one would have to do [2ND] [recall] in order to take a look at variable values. A definite minus for me. But overall my verdict is that the Mathview feature alone places this device at the forefront of calculators in the $20 and beneath range.

Now if Texas Instruments would finally freaking release a calculator that will give accurate z-scores and CDF's of the standard normal distribution!! :judge:

uclatommy

10-02-2009, 11:25 PM

CC, you don't have to store stuff in memory. You can move your cursor back through the history and pick out stuff you want to reuse. Also, I've never used the fraction function on a calculator until this one. Its pretty awesome. The numerators and denominators don't need to be a single number. You can input a calculation in each.

badmaj5

10-03-2009, 10:11 AM

Got this yesterday.

You can just hit [2ND] [Left] or [Right]

I did not know that, thanks for the tip!

volva yet

10-03-2009, 09:56 PM

Why can't we just use Excel on exams?:shrug:

uclatommy

10-04-2009, 05:43 PM

Why can't we just use Excel on exams?:shrug:

The data key is a baby spreadsheet. You have 3 columns and the columns can be functions of the values in another column. You can then use the stat function on a column to find sums, averages, and variances.

This feature really shines on SOA #144. I can solve this problem very quickly with this calculator. First, find E[(X-100)]/E[X] for each of the 10 simulations. You can do this in your head pretty easily. Next, input these values into column L1. In L2, use the formula ((1-L1)-0.125)^2. Then go to stat and find the average of L2 and that is your answer.

JavaGeek

10-05-2009, 03:06 PM

One thing I love to do with integrals is to use a approximating technique to solve rather than say integration by part (which I always make small mistakes on). Table mode allows me to get a lot of function results without having to edit the values all the time so I can find

\int_0^{\infty}\frac{x^3e^{-x/100}}{100^3}

Using "simpson's rule (http://en.wikipedia.org/wiki/Simpson%27s_rule)":

step size = 125

Go from 0 to 1250.

\frac{125\times 2}{6}(0 + 4\times 0.560 + 2\times 1.28 + 4\times 1.24 2\times 0.842 + 4\times 0.471 + 2\times 0.233 + 4\times 0.106 +2\times 0.0454 + 4\times 0.0185 + 0.00727)=599

The rest of the integral is essentially 0:

\int_{1250}^{\infty}\frac{x^3e^{-x/100}}{100^3}=0.932

True answer = 600...

note: use the multi-line calc to get the values and a backup calculator to apply the simpon's rule

Of course you have to make sure you're spacing such that you're going through about 10% of the function dominant area each step... If your steps are too large you'll get the wrong answer... [For example a step size of 156.25 instead would get an answer of 608, step size=208.3 would give 626]

Of course the solution to the above integral should be memorized...

\frac{3!100^4}{100^3}=600

uclatommy

10-05-2009, 06:49 PM

One thing I love to do with integrals is to use a approximating technique to solve rather than say integration by part (which I always make small mistakes on). Table mode allows me to get a lot of function results without having to edit the values all the time so I can find

\int_0^{\infty}\frac{x^3e^{-x/100}}{100^3}

Using "simpson's rule (http://en.wikipedia.org/wiki/Simpson%27s_rule)":

step size = 125

Go from 0 to 1250.

\frac{125\times 2}{6}(0 + 4\times 0.560 + 2\times 1.28 + 4\times 1.24 2\times 0.842 + 4\times 0.471 + 2\times 0.233 + 4\times 0.106 +2\times 0.0454 + 4\times 0.0185 + 0.00727)=599

The rest of the integral is essentially 0:

\int_{1250}^{\infty}\frac{x^3e^{-x/100}}{100^3}=0.932

True answer = 600...

note: use the multi-line calc to get the values and a backup calculator to apply the simpon's rule

Of course you have to make sure you're spacing such that you're going through about 10% of the function dominant area each step... If your steps are too large you'll get the wrong answer... [For example a step size of 156.25 instead would get an answer of 608, step size=208.3 would give 626]

Of course the solution to the above integral should be memorized...

\frac{3!100^4}{100^3}=600

Great idea using numerical methods to approximate an integral, but we can do better. Rather than relying on a second calculator, we can do this all with the data key and the 3 columns. Here's how:

1. In column L1, manual input the sequence 1,2,3,...,n. To make this faster, you can press [1][enter], [+][1][enter], [+][1][enter],...

2. In column L2, use the following formula: sin(L1*90)^2. [enter] This gives the sequence 1,0,1,0,... This will be used to generate the coefficients 4,2,4,2,...

3. In column L3, use the following formula: (2+2*L2)(0.01^3)((125*L1)^3)(e^(-1.25*L1)). [enter]

4. [2nd][quit] and get the sum of column L3 from the stat menu.

5. Multiply by 125/3.

When I use 15 steps, I get an answer of 600.83. If you go out far enough, the last term f(x_n) won't really matter. You should still calculate f(x_0) and add it to the sum, but I skipped that step knowing that f(x_0) would be 0.

Also, to help determine the step size of Simpson's rule, we can use the table key to get an initial idea of how the function behaves and what it's domain is.

Additionally, instead of using the fixed value 125 in L3, we can store the step size into a variable. (i.e. 125 [sto->] x)

So L3 now becomes (2+2*L2)(0.01^3)((x*L1)^3)(e^(-x*.01*L1))

Now we can futz around with the step size and expand the number of steps and everything will auto-calculate. When I use step size 100, and 20 steps, I get 599.47.

JavaGeek

10-05-2009, 07:11 PM

Great idea, that would prevent silly typo's

However, it might be easier just to type L2=1,4,2,4,2,4,2...4,1 and then calc:

1-VarStats Stat = L1, Freq = L2. Get \frac{\mbox{step size}\time2}{6}\sum X

uclatommy

10-05-2009, 07:24 PM

Great idea, that would prevent silly typo's

However, it might be easier just to type L2=1,4,2,4,2,4,2...4,1 and then calc:

1-VarStats Stat = L1, Freq = L2. Get \frac{\mbox{step size}\time2}{6}\sum X

You're right. This seems more efficient and less prone to invalid functions.

JavaGeek

10-05-2009, 07:33 PM

The easiest way to do the numerical integration is:

type in the X values you want to use in L1 (find out what function looks like using table)

type in the seq 1,4,2,4,2,...2,4,1 into L2

click [data] -> goto column L3 [data] -> Formula -> Add Edit Frmla

Type function, when you need X click [data] -> L1

Eg.

x^2*exp(-X): [data][enter][x^2] [2nd][e^x][(-)][data][enter][)]

When done hit [enter]

It will have to think for 30 seconds or so.

Then go to [2nd][stat] -> 1-Var Stats -> Data = L3; FRQ = L2; CALC [enter]

If you try to put the function into L2 and then do the sequence (1,4...1) it will be really slow. So always do the function last.

Note: some comments about the memory - the slowness of the data table is likely the result of a slow CPU and not poor memory - it appears they are refreshing the entire table each time you scroll (which is stupid)

uclatommy

10-05-2009, 07:55 PM

I prefer using the sequence 0,1,2,3... in L1.

I use 125 [sto->] x before going into data mode. This will let me change step sizes easily.

The evaluation points used in L3 are:

L1*x + a

where a is the starting point of the integration. In the example, a=0.

Now I can test other step sizes to see to find one that is small enough to produce a good estimate.

Also, in L3, I multiply everything by x/3 so the stat function will give the answer without any additional calculation.

actuarialmath

10-06-2009, 04:42 PM

Can we use this calculator to do the matrix calculations?

badmaj5

10-06-2009, 04:53 PM

What matrix calculations are you talking about? The only matrix operation I can think of is inverting the information matrix, and that isn't anything that should require fancy calculator techniques to get in the 2x2 case.

volva yet

10-06-2009, 07:25 PM

My coworker has this calculator. I am definitely going to buy it.

Coquelicot

10-07-2009, 08:25 AM

I just got this calculator last night...SO worth the $15.....amazing! :-) Thanks to the OP for bringing up how great it is!

actuarialmath

10-07-2009, 03:14 PM

What matrix calculations are you talking about? The only matrix operation I can think of is inverting the information matrix, and that isn't anything that should require fancy calculator techniques to get in the 2x2 case.

Delta method could also use matrix calculation.

badmaj5

10-07-2009, 03:25 PM

True. When there are 2 parameters though, I find it easier to just use the delta method on each parameter and include a covariance term than deal with the matrices.

chris_arwood_sucks

10-09-2009, 07:17 PM

Why is this calculator allowed on exams?

badmaj5

10-09-2009, 07:23 PM

Why is this calculator allowed on exams?

Because regardless of what calculator you used, the last exam was ridiculous.

no driver

10-13-2009, 10:02 AM

Ok, say we are looking at SOA problem # 172, where we want to do a K-S test with the null hypothesis that F(x)=1-\frac{1}{(1+x)^4} and our data values are 0.2, 0.7, 0.9, 1.1, and 1.3. On the Multiview, you can push the Table button, it asks for y as a function of x, so you type in y=F(x), and then the next thing that comes up is it asks for the increments. Set that to manual, and then on the table you just type in the 5 x values and it tells you the 5 values of F(x). There is an old exam problem with 10 data values, so you save even more time there. You can do the same thing with the inversion method.

The stat functions are also huge, and you should know how to find the unbiased estimate of the variance of weighted data on your calculator, but that can be done on the 30XIIS as well as the Multiview.

For those of you in my online seminar, I will try to put together a calculator tips video at some point late next week.

Heads up for TIA folks, Dave posted his lesson on the TI-30XS last night.

Coquelicot

10-13-2009, 12:54 PM

Heads up for TIA folks, Dave posted his lesson on the TI-30XS last night.

sa-WEET!

racerx

10-20-2009, 07:36 PM

I saw Dave's instructions on the TI-30 MultiView. It looks awesome. Do you think it is too late to get used to the calculator before Nov. 9?

Thanks

chopsticks

10-20-2009, 07:53 PM

I sure hope not. Mine is supposed to be delivered tomorrow, so we'll see. I am already pretty familiar with the TI-30, and it is very similar to that one. The biggest thing will be learning all of the tricks mentioned in this thread, and remembering them on the day of the exam.

colemstpeter

10-21-2009, 06:45 AM

I'm going to put all my TI-30XS calculator tricks in this thread. If you see anything that could be done better or you've got a different method of approaching a problem please share. Eventually I'll be glad to summarize all of these tricks (as well as those submitted by other posters) and add them to the sticky.

Instead of entering a long equation that might include typos, I use the [data] lists to calculate and sum my Chi-square test statistic.

1) [data]->L1 enter your expected occurrences

2) L2 enter your observed occurrences

3) With your cursor in L3, [data]->FORMULA->Add/Edit Frmla

4) Enter: ([data]->L2 - [data]->L1)^2/[data]->L1. When you are done it should look like L3=(L2-L1)^2/L1

5) [2nd][stat]->1-Var Stats->DATA=L3 FRQ=ONE->CALC

6) Find ∑x and compare it to the Chi-square table

If the problem only has a few groups of data it might be faster to do it by hand but on a problem with several groups of data you'd have to break the sum into multiple chunks to enter it into the calculator anyway and there's more chance to make a mistake with missed parentheses etc.

With this method if you notice a typo in your entered data in L1 or L2 you can correct that entry and it will recalculate the corresponding value of L3 automatically. You'll still need to recalc the stats to get the new sum of L3.

I started doing this and now the only time I pick up my pen for a Chi-Square problem is to mark the answer. Thanks!!!

daaaave

10-21-2009, 09:29 AM

I saw Dave's instructions on the TI-30 MultiView. It looks awesome. Do you think it is too late to get used to the calculator before Nov. 9?

Thanks

If you are already using an earlier TI-30, then it doesn't take long to get used to the new layout of things. If you are converting from a BA-II+, it takes longer to get used to the different order of keystrokes with ln() and e^(), but if the MultiView is the only calculator you use for 2 weeks then you should be fairly familiar with it by the end.

Coquelicot

10-21-2009, 10:11 AM

If you are already using an earlier TI-30, then it doesn't take long to get used to the new layout of things. If you are converting from a BA-II+, it takes longer to get used to the different order of keystrokes with ln() and e^(), but if the MultiView is the only calculator you use for 2 weeks then you should be fairly familiar with it by the end.

This was my situation - I got it down within a few hours of using the thing pretty easily. I still slip up on ln() and e^() from time to time, but you can see it if you mess it up, so that's a huge benefit and it's super-easy to go back and fix. There are so many benefits to this calculator and it is so cheap -- thanks to OP for mentioning how awesome it is! It was a suprisingly easy switch.

concactu

10-21-2009, 01:19 PM

I bought this calculator about a year ago, but quickly discarded it after a month. It was way too slow compared to the earlier TI-30, and as a result I kept on getting wrong answers cause it wasnt recognizing all my key entries.

As well, it is beyond my understanding why they moved the '+' key up one and stuck that other key in its place (can't recall the name). After years of using calculators, I found it too difficult to adjust to the new placement of the + key.

That being said, it has great features. Some of the tips mentioned above sound great, and I'm looking forward to trying them out. But the earlier TI-30 will always be my primary calc.

racerx

10-21-2009, 03:44 PM

I got the TI-30xs Multiview and it is great. I would recommend getting this calculator now so that you can be familar on the exam. I am a little resistant to new technology and found it easy to use.

Of course I have TIA so I got the video lessons to help me learn it - Thanks Daaaave.

I just wish it came in pretty colors like the Ti-30XIIS. Oh well we can't have everything.

Back to :study::study:

actuwriter

10-23-2009, 12:08 AM

does anybody know if the TI 30XS has a comma separator option? I love on the BAII Plus that you can easily see that you have typed in 60,000,000. You can easily miss a zero on the TI 30's.

GooseyGoose

10-23-2009, 12:36 AM

does anybody know if the TI 30XS has a comma separator option? I love on the BAII Plus that you can easily see that you have typed in 60,000,000. You can easily miss a zero on the TI 30's.

I scanned the manual and didn't see an option. There is a comma on the 2nd function of the ".", but that's used for separating values in formulas and stuff. I have the same issue as you, and I'm going to get used to using the x10^n key for values like 63 million so I don't miss a zero.

badmaj5

10-23-2009, 09:21 AM

It sounds really stupid, but I count zeros in sets of 3. So when I have to write something like 63,000,000 in my head I'm thinking "63...one two three...one two three". Then again, using the memory is a good choice I think for keeping these values throughout the entire problem and then not having to worry about typing anything out.

Clam Chowdah

10-23-2009, 09:56 AM

CC, you don't have to store stuff in memory. You can move your cursor back through the history and pick out stuff you want to reuse. Also, I've never used the fraction function on a calculator until this one. Its pretty awesome. The numerators and denominators don't need to be a single number. You can input a calculation in each.

True but then the input prompt has less space for you to punch in your calculations.

I'm going to put all my TI-30XS calculator tricks in this thread. If you see anything that could be done better or you've got a different method of approaching a problem please share. Eventually I'll be glad to summarize all of these tricks (as well as those submitted by other posters) and add them to the sticky.

Instead of entering a long equation that might include typos, I use the [data] lists to calculate and sum my Chi-square test statistic.

1) [data]->L1 enter your expected occurrences

2) L2 enter your observed occurrences

3) With your cursor in L3, [data]->FORMULA->Add/Edit Frmla

4) Enter: ([data]->L2 - [data]->L1)^2/[data]->L1. When you are done it should look like L3=(L2-L1)^2/L1

5) [2nd][stat]->1-Var Stats->DATA=L3 FRQ=ONE->CALC

6) Find ∑x and compare it to the Chi-square table

If the problem only has a few groups of data it might be faster to do it by hand but on a problem with several groups of data you'd have to break the sum into multiple chunks to enter it into the calculator anyway and there's more chance to make a mistake with missed parentheses etc.

With this method if you notice a typo in your entered data in L1 or L2 you can correct that entry and it will recalculate the corresponding value of L3 automatically. You'll still need to recalc the stats to get the new sum of L3.

Just saw this and I am blown away. Along with speeding up statistical calculations(finding variance, covariance and critical values for K-S, Chisq, etc.), The TI-30XS almost completely solves the problem of errors in data input which already compose about 20% of the practice problems I get wrong. I take back everything I said about this calculator being inferior.

volva yet

10-25-2009, 05:10 PM

Kaplan-Meier and Nelson-Aalen estimates deserve to be used by this calculator but they require several more columns. Ideas anyone? BTW, I bought the calculator yesterday and love it so far!

uclatommy

10-25-2009, 07:21 PM

You can clear formulas and it will do a "paste values" in every column. Then you can erase columns without worrying about messing up a previously calculated column. It's a good way to continue calculation in a column that you dont need anymore.

Clam Chowdah

10-28-2009, 11:12 PM

From the table function is there a way to store y-values into a variable? This would speed up, Anderson-Darling calculations, Kolmogorov-Smirnov calculations...

volva yet

10-28-2009, 11:20 PM

From the table function is there a way to store y-values into a variable? This would speed up, Anderson-Darling calculations, Kolmogorov-Smirnov calculations...

I recommend plugging your formulas in to the data section - the mini spreadsheet that you can run statistics on.

It is limited to three columns, but once you are done with a column (nothing else further will depend directly on it) then you may clear the formula. Do not delete it, just click your data button, go over one to the right to the formula section, and pick #2, #3, or #4. What that does is essentially "range value" (Excel lingo) your formula results - ya know, copy + paste special + value :wink:.

IAHAWKEYE

10-29-2009, 11:57 AM

Just got this calculator yesterday, a question:

I entered data into a list and then used the 1-Var Stats option to calculate n, x bar and so on. My question is: Is Sx= unbiased s.d, and sigma(x) = biased s.d.? This seems backwards to me, am I wrong?

badmaj5

10-29-2009, 12:01 PM

I've always seen S denote the unbiased standard deviation. When you find the number of simulation runs needed, or you're working with semi-parametric bayesian analysis, you use S and work with unbiased variance/standard deviation.

daaaave

10-29-2009, 12:05 PM

If you are given both numbers, you can always remember which is the unbiased as the unbiased sd is > the biased sd because we are dividing by a smaller denominator.

IAHAWKEYE

10-29-2009, 12:31 PM

If you are given both numbers, you can always remember which is the unbiased as the unbiased sd is > the biased sd because we are dividing by a smaller denominator.

True. Thanks.

success_will_write

11-01-2009, 12:16 PM

$14.96 well spent at WalMart

finally watched Dave's calculator lecture

probably should have done this several weeks ago

volva yet

11-01-2009, 01:58 PM

finally watched Dave's calculator lecture

probably should have done this several weeks ago

Is that open to everyone? If so... link please?

actuarialmath

11-02-2009, 08:33 AM

Has anyone had this problem with TI-30XS Multiview?

For the STAT function, you select 1-Var stats, DATA:L1; FRQ:ONE and hit CALC. It will give you the answer screen: 1-Var: L4,One and all the wrong answers.

On the DATA function, there are only 3 columns: L1,L2,L3. How come the STAT answer is based on L4?

volva yet

11-02-2009, 11:00 AM

Has anyone had this problem with TI-30XS Multiview?

For the STAT function, you select 1-Var stats, DATA:L1; FRQ:ONE and hit CALC. It will give you the answer screen: 1-Var: L4,One and all the wrong answers.

On the DATA function, there are only 3 columns: L1,L2,L3. How come the STAT answer is based on L4?

That's strange; I've never seen that! :slug:

no driver

11-02-2009, 11:38 AM

I frequently notice garbage in my first entry of L1 after I am out of the DATA lists and then come back. Usually I leave the DATA lists to do STAT calculations, so maybe this is the cause, but I haven't pinned down the behavior enough to know for sure.

It doesn't seem to affect the stat calculations, and formulas in the other lists that depend on L1 don't seem to have changed, but when I go back into my DATA lists the first entry is corrupted. I'm usually ready to move on to another problem when I notice it so I clear the lists anyway. I'm not sure what would happen if it were necessary to do additional work with L1, but it's something to be aware of.

Maybe this isn't a problem for other folks, but thought I'd mention it just in case.

concactu

11-02-2009, 12:17 PM

I frequently notice garbage in my first entry of L1 after I am out of the DATA lists and then come back. Usually I leave the DATA lists to do STAT calculations, so maybe this is the cause, but I haven't pinned down the behavior enough to know for sure.

It doesn't seem to affect the stat calculations, and formulas in the other lists that depend on L1 don't seem to have changed, but when I go back into my DATA lists the first entry is corrupted. I'm usually ready to move on to another problem when I notice it so I clear the lists anyway. I'm not sure what would happen if it were necessary to do additional work with L1, but it's something to be aware of.

Maybe this isn't a problem for other folks, but thought I'd mention it just in case.

I have seen the same thing.

Clam Chowdah

11-05-2009, 09:31 PM

ASM 9TH Edition Lesson 50 is entitled "Bühlmann As Least Squares Estimate of Bayes"

You can blaze through the problems requiring you to find the coefficients "a" and "b" Go to Data and input in [L1] the independent variable: the raw data. In [L2] input the Bayesian estimates. Of course you have to account for the proportions that each unique data value takes up the set. Say the ordered pair (X_i,Y_i) is a raw datum & Bayesian estimate, respectively. If you are given ten ordered pairs of (5,7) and five ordered pairs of (7,9) you only have to input two rows of the former and one row of the latter.

Then you can go to Stat and 2-var stats and calculate the Bühlmann credibility "b" and the "a" value.

IMPORTANT!!

The number given by the MultiView labeled "b" is actually the a value -- the y-intercept and the the number given by the MultiView labeled "a" is actually the Bühlmann credibility Z or b. Obviously if the value of a displayed on the screen is greater than 1 you can deduce which entity is NOT the Bühlmann credibility.

chris_arwood_sucks

11-07-2009, 01:11 PM

Oh my god, I just got this thing. So superiour to the TI-30XIIS. This table function is so amazing. Why did I not get this sooner. This will eliminate so many calculator mistakes because you're able to see so much more of what you are typing in.

The only thing that could beat this is illegal calculators.

chris_arwood_sucks

11-07-2009, 01:25 PM

I just bought this thing and am taking the test on Monday. Is that a stupid idea (obviously waiting this long was a stupid idea)?

I figure I'll bring this beast along with my more-familiar TI-30XIIS and at the very least utilize the 'table' option on the Multiview.

Is there any advantage to using the STAT fcn on the Multiview vs. the TI-30XIIS? Skimming throught this post, sounds like some people have had problems with it.

daaaave

11-07-2009, 01:39 PM

I think that the way you input data is more intuitive on the MutiView, which is why I prefer doing stats on the MultiView. My recollection is that you get the same outputs with both (or at least the same key outputs that we care about), but I haven't used the 30XIIS recently.

chris_arwood_sucks

11-07-2009, 01:57 PM

Question:

In STAT, what, if anything, can you do with the "FORMULA"? It seems promising but I can't seem to enter anything without causing an error.

chris_arwood_sucks

11-07-2009, 01:59 PM

I think that the way you input data is more intuitive on the MutiView, which is why I prefer doing stats on the MultiView. My recollection is that you get the same outputs with both (or at least the same key outputs that we care about), but I haven't used the 30XIIS recently.

Thanks Dave, I compared the two as far as what I would regularly do with the 30XIIS, and they seem to yield identical results. Although the user's manual is impressively not-all-that-helpful. Which is why I try to never read them.

JavaGeek

11-07-2009, 02:11 PM

Question:

In STAT, what, if anything, can you do with the "FORMULA"? It seems promising but I can't seem to enter anything without causing an error.

You have to "grab" other column values by pressing "[DATA]" again and picking L1, L2 or L3, depending on what you want as an input into the function.

chris_arwood_sucks

11-07-2009, 05:49 PM

You have to "grab" other column values by pressing "[DATA]" again and picking L1, L2 or L3, depending on what you want as an input into the function.

So I can grab the L1 vector and perform the same function on all L1 values to populate my L3 vector?

I hope that's what you're saying, because that will just work out swell if I have to use the invariance principal for MLE solving.

rahim

11-08-2009, 02:29 AM

em now going for C in may . My frnd has got ti-30xIIS but em not sure wat's best for C's statistical functions. ryt now i got BAII plus which also got some "STAT" keys . Wat do u guys prefer me to buy ? ti-30X IIS or ti-30x multiview or to remain with my BAII plus?

daaaave

11-08-2009, 08:40 AM

The multiview. There are a number of topics that require somewhat tedious calculations that are much faster on the multiview.

Hawgdriver

11-11-2009, 08:36 PM

Kaplan-Meier and Nelson-Aalen estimates deserve to be used by this calculator but they require several more columns. Ideas anyone? BTW, I bought the calculator yesterday and love it so far!

use paper for determining the r's and the x's, then input those into L1 and L2

from here, you can use L3 and formulas for KM and NA (just use ln (S(x_{j})) instead of Sj, so that you can use the stat function to get \sum(S(x_j)) which you can easily convert to an overall survival until event j (just delete the final Sj's before you get the sum--you will have to kill the formula in L3 to delete entries. to do this, I usually just re-enter the top number, but you can do it using the menus with fewer keys). NA, variance, and other calcs should be self-expanatory. The key for efficiency is to put the "r" in L1 (or the P, if UDD) and the "x" in L2. You can get all the KS/NA formulas quickly this way. Just derive the r's and x's on paper w/ a table, especially if it's not sorted or there are d's and c's.

Hawgdriver

11-11-2009, 08:40 PM

I can't believe they allow this calculator. Anyone not using it is behind the curve. The table view is nice, but the data function is the real money-maker in my book. I use it for any problem with data. Empirical credibility, bootstrap, KS, chi, MLE, simulation, etc. There is a problem solving algorithm using this calculator that is faster than without it for any problem with a list of data.

Oh, and thanks for the tip. Thread of the year.

akfisherman

11-12-2009, 12:22 AM

i discovered this thread and purchased the calculator less than a week before my exam. definitely worth the 15 bucks. :clap:

Hawgdriver

11-14-2009, 09:14 PM

Here is how I used the Multi-view's data and stats functions to solve a Bayes problem.

The problem is in ASM (8th ed?) #45.26, the first bayes chapter. It was on the F00 course 4 exam, problem #28 (page 29) (http://www.soa.org/files/pdf/course4_1100.pdf).

I will show two methods for solving the problem, the first is a standard approach and the second takes advantage of the "data" function using multiview (MV).

The problem:

There are three classes of risks, each equally likely. They follow a pareto distribution with \theta=10.

1: \alpha=1

2: \alpha=2

3: \alpha=3

There is an observation of 20.

What is the probability of the next observation being greater than 30?

Method 1:

I timed this. It took me three minutes. I had already solved it twice, so this was a pretty speedy solution. I tried to go through the method like I would for an exam, but knowing the facts and method and never stalling to "think" about how to solve it. Just pure algorithm...if that makes sense.

To solve:

Make a table with three column headers, 1, 2, 3 for the values of \alpha.

put 1/3 under each.

compute the value of the probability density function

f(x)=\frac{\alpha\theta^{\alpha}}{(x+\theta)^{a+1} },\;\theta=10

for each at x=20. Put the result in the appropriate column. (.0111, .0074, .0037)

sum the results (0.0222)

divide each joint probability by the sum to find the posterior probability (.5, .333, .167)

compute the value of S(30) for each: (.25, .0625, .0156)

multiply the posteriors by the corresponding S(30) and take the sum: 0.148

Method 2, using MV:

In L1 put 1, 2, 3

In L2 put in the formula L2=L1 10^L1/30^(L1+1 {you don't need a space or "times" between L1 and 10, and you don't need a final ")" on expressions.

[2nd] [mode] {quit}

[2nd] [data] {stat}

1Var stats, L2 data, hit [5] to return \sum{x} then [sto>] [x]

[data] enter a formula into L3: L2/x

"freeze" L3 by re-entering .5 into the top cell (if this step confuses you, try to skip it and see the result)

enter a formula into L2: .25^L1

"freeze" L2 by re-entering .25 into the top cell

enter a formula into L1: L2L3 (you don't need to hit the multiply button)

[2nd] [mode] {quit}

[2nd] [data] {stat}

1Var stats, L1 data, hit [5] to return \sum{x}, which is 0.148

This took me two minutes, but I made a few mistakes and had to start over. The main time savings is due to the fact you don't have to compute 3x each of (density, posterior, S(30)). I'm convinced it's worth the effort to make the [data] function and the [stat] function your friend. Also less likelihood of error.

If method 2 seems weird, it will seem natural with a little practice.

Anyone disagree? Dissent is welcome.

:bump:

Since the calculator shortcuts are a hot topic in the other thread, I thought that August sitters might find reading through this thread useful. Thanks to everyone who posted tips and tricks previously.

yashi911

07-26-2010, 08:19 PM

Va va

yashi911

07-26-2010, 10:05 PM

Does this calculator do integrals?

Spiritbreaker

07-26-2010, 10:18 PM

Nope it doesn't.

love for math

07-27-2010, 07:25 PM

oh my god, this calculator is ridiculous. the worst part is that i've had it for a while now (about a year) and had no idea what i was working with. i really only liked it because you could scroll up to other values, and see the entire formula you were typing.. little did i know.... wow, thanks guys!

arista

07-27-2010, 08:59 PM

Just purchased this calculator. Lets hope I can master it's capabilities in 2 weeks!

Hawgdriver

07-30-2010, 12:26 AM

Does this calculator do integrals?

I think JGeek has a post in this thread describing how he uses a numerical technique for this. I haven't used it, but it might be perfect for certain problems. It might also be nice to have more than one, I've found that for some problems I don't know the right way to solve them, so I might use the data function and trial and error, but then need to clear the data to use it for another problem. Having two would remedy this.

You can bring as many of calculators as you wish. The TI business calculator has the IRR function that you can use to quickly solve the quadratic formula*. So you get the best of both worlds.

*use CF0, 1, and 2 to set up a quadratic in the form of x=1/r. IRR gives you one solution instantly (remember it's a percentage), and the other is ANS^-1 * C / A (where C is CF0 and A is CF2). Thank Guo for that one.

yashi911

07-30-2010, 12:50 AM

I think JGeek has a post in this thread describing how he uses a numerical technique for this. I haven't used it, but it might be perfect for certain problems. It might also be nice to have more than one, I've found that for some problems I don't know the right way to solve them, so I might use the data function and trial and error, but then need to clear the data to use it for another problem. Having two would remedy this.

You can bring as many of calculators as you wish. The TI business calculator has the IRR function that you can use to quickly solve the quadratic formula*. So you get the best of both worlds.

*use CF0, 1, and 2 to set up a quadratic in the form of x=1/r. IRR gives you one solution instantly (remember it's a percentage), and the other is ANS^-1 * C / A (where C is CF0 and A is CF2). Thank Guo for that one.

Is it X= 1/r or X=1/(1+r)

Also thanks for the info, i saw javageeks post about approximating integrals, its a bit too much for me, and i dont really understand it! Anyways im okay enough at calc to sqeak it through.

nxtMtl

07-30-2010, 10:40 AM

I frequently notice garbage in my first entry of L1 after I am out of the DATA lists and then come back. Usually I leave the DATA lists to do STAT calculations, so maybe this is the cause, but I haven't pinned down the behavior enough to know for sure.

It doesn't seem to affect the stat calculations, and formulas in the other lists that depend on L1 don't seem to have changed, but when I go back into my DATA lists the first entry is corrupted. I'm usually ready to move on to another problem when I notice it so I clear the lists anyway. I'm not sure what would happen if it were necessary to do additional work with L1, but it's something to be aware of.

Maybe this isn't a problem for other folks, but thought I'd mention it just in case.

Hi All:

As per above, I also have often noticed corruption in L1 when returning to the data table. The numbers are getting corrupted somehow but I am unable to come up with the exact sequence that causes this problem.

Perhaps it is an intermitant problem ?

Has everyone on this thread noticed this problem ? Perhaps it is only affecting a certain production run of the calculator ?

There is no mention of this bug on the TI website.

This definitely does not give me a "warm fuzzy feeling" even though this calculator is very powerful. I just really hope there are no other "bugs" in this calculator that are occuring but which are less easy to notice !

nxtMtl

yashi911

07-30-2010, 11:34 AM

I havent had this problem, the only time L3 freezes for me is when i have a formula there but no data in l2 or L1

Hawgdriver

07-30-2010, 01:05 PM

Is it X= 1/r or X=1/(1+r)

Also thanks for the info, i saw javageeks post about approximating integrals, its a bit too much for me, and i dont really understand it! Anyways im okay enough at calc to sqeak it through.

Good catch, 1/(1+r).

yashi911

07-30-2010, 01:50 PM

Good catch, 1/(1+r).

i wasnt sure, ive pushed all my FM skillz out of my head for this exam!

Hawgdriver

07-30-2010, 02:03 PM

i wasnt sure, ive pushed all my FM skillz out of my head for this exam!

I picked that one up in Guo's little manual that should be called 'possibly useful shortcuts and some tedious proofs'. For whatever reason, it takes me forever to solve the quadratic formula with the multi-view.

Maybe what I should do is enter the formulas

(-Y + (Y^2 -4XZ)^.5)/(2X)

(-Y - (Y^2 -4XZ)^.5)/(2X)

into my 'backup' calculator during the instruction screen time and then just scroll up to it if I need it. It seems there's always an MLE or some other problem that boils down to the QF.

This is probably overkill.

JSA1987

08-07-2010, 10:57 PM

Hi All:

As per above, I also have often noticed corruption in L1 when returning to the data table. The numbers are getting corrupted somehow but I am unable to come up with the exact sequence that causes this problem.

Perhaps it is an intermitant problem ?

Has everyone on this thread noticed this problem ? Perhaps it is only affecting a certain production run of the calculator ?

There is no mention of this bug on the TI website.

This definitely does not give me a "warm fuzzy feeling" even though this calculator is very powerful. I just really hope there are no other "bugs" in this calculator that are occuring but which are less easy to notice !

nxtMtl

I have this issue as well.

For me, it is strictly in cell L1(1) -- i.e., column A, row 1.

Hawgdriver

08-09-2010, 05:23 PM

Does this calculator do integrals?

Not sure if this is a verbatim repeat of JavaGeek's numerical method, but I ended up using the DATA function for an integration during a practice exam. I saw that I had to find the value of a gamma function to determine the S(2.5) for an inverse gamma, and there was no way I was going to solve this by hand:

(1/2) t^2 e^(-t) from 0 to 1.6 (http://www.wolframalpha.com/input/?i=integrate+%28.5t^2+e^%28-t%29%2C+0%2C1.6%29)

i.e. gamma(3,1.6)

So it came to me that I could use the DATA function to solve it.

I filled in 0 thru 16 on L1 <--this is the range of integration

used formula .5*(L1/10)^2 e^(-L1/10) in L2 <--this is the density function

copied L2 into L3, deleted the top value, deleted the bottom value of L2

created the 16 rectangle-areas with the formula: .05 (L2 + L3) <-- that's 0.1 (interval width) times (1/2 to average L2 and L3)

used STAT to find the sum of X1, that's the approximation

It was more accurate than I expected, .21669 compared to the actual value of .21664

Now, it turns out that I was solving a different problem than the one I was required to solve on the practice exam. I didn't *really* think the exam would test the ability to integrate t^2 e^-t. But who knows if it will come in handy some time.

This is probably the same thing already posted above, but maybe this helps someone else.

JSA1987

08-12-2010, 09:44 PM

I just read this feature of the calculator on the TI website:

"Battery powered with solar cell assistance to lengthen battery life"

This makes me a little nervous. I think I'd better change my battery before Monday. Or just buy a new calculator, since I'm having the issue with corruption in the data table.

yashi911

08-12-2010, 09:52 PM

I just read this feature of the calculator on the TI website:

"Battery powered with solar cell assistance to lengthen battery life"

This makes me a little nervous. I think I'd better change my battery before Monday. Or just buy a new calculator, since I'm having the issue with corruption in the data table.

get a new one, the battery will cost you 10$ anyways. I have never had this corruption of data problem. Try resetting you calc perhaps.

JSA1987

08-12-2010, 10:30 PM

Yeah, I'll probably buy a new one and use this one as a backup. It's a little annoying because I had an extra and I can't find it, but I guess that is water under the bridge. Certainly it's worth having the new one for $15 or whatever it is.

Hawgdriver

08-12-2010, 10:52 PM

I've been integrating more often now with DATA. Is anyone interested?

*crickets*

big picture -- technique is that you use only ten intervals. If the function is over an infinite interval, or the function is squirrely, you could split the interval into two segments and run through this process twice (well, just the last two steps).

clear DATA

enter 1 thru 10 in L1

enter this formula

L1-.5

into L2 <-- so it takes the halfway point value of the interval

set L3 equal to L2 <-- we use this later to scale it to the interval of integration

set L1 equal to the function that is integrated in terms of L3, so y(x) is L1(L3)

go back and use L3 as a 'scale' function of the interval in L2, for example, if the interval is [4, 25] we need to stretch our [0,10] in L2, and then translate it laterally. The formula for L3 then becomes L3 = 21/10 L2 +4 (25-4 = 21, so 21/10 * L2 is the scale transformation, plus the beginning of the interval -- 4. Now L3 is a set of ten interval midpoints over the length of the interval of integration-- { 5.05, 7.15, 9.25, ... , 19.95 }) You could do this step earlier if you want. I don't.

pull the sum of the rectangle lengths from STAT and multiply it by the interval size (2.1) to get the approximation

this method is fast because you can break a complex interval into chunks by rescaling L3 a couple of times.

I used this for the Anderson Darling problem on one of the ASM practice exams because I wasn't smart enough to use a calculus trick (and didn't know the long ass formula and refuse to memorize that one). My calculus skills aren't what they were 18 years ago...

maybe this sounds time consuming. well, if you can't do an integral in closed form, it's faster than infinity. then once you do it a couple times, you do it fast and learn to lean on it... I've been doing it a lot now.

yashi911

08-12-2010, 11:46 PM

U crazy hawg driver

childplease

08-12-2010, 11:49 PM

thats insane. its just integration by parts is all nothing too ground breaking

yashi911

08-14-2010, 08:40 PM

So the table function can also be used for MLE's.

Taking a relatively tedious example like ASM PE5 #5 where we have to Maximize the product of exponentials and poisson.

1. Go through and build the Likelyhood function, you should end up with ; (e^((-15(x^2)-8)/3x))/(162)

This is the compressed version of the Likelyhood, you dont have to simplify, but i think it would be harder to input a function with so many parentheses.

----> Hit table then you will get the y= prompt. Input the function. Then look at the possible answers, start = .5 step = .01 , hit okay. You will get the x and y columns, now follow the y column and look for where y stops increasing and begins to decrease, This happens at x = .73 , and that is your answer!

Becareful of decimal places make sure you you can see all the relevant ones. First time i went through i was only looking at 2 sigfigs, make sure you look at as many as possible so you can come up with the correct answer.

I think it can be applied to any MLE as long as you get the correct Likelyhood function. Hope this helps alleviate some of the tedium involved in MLE's.

Ps i think they are eventually gonna ban this calc!

childplease

08-14-2010, 09:33 PM

So the table function can also be used for MLE's.

Taking a relatively tedious example like ASM PE5 #5 where we have to Maximize the product of exponentials and poisson.

1. Go through and build the Likelyhood function, you should end up with ; (e^((-15(x^2)-8)/3x))/(162)

This is the compressed version of the Likelyhood, you dont have to simplify, but i think it would be harder to input a function with so many parentheses.

----> Hit table then you will get the y= prompt. Input the function. Then look at the possible answers, start = .5 step = .01 , hit okay. You will get the x and y columns, now follow the y column and look for where y stops increasing and begins to decrease, This happens at x = .73 , and that is your answer!

Becareful of decimal places make sure you you can see all the relevant ones. First time i went through i was only looking at 2 sigfigs, make sure you look at as many as possible so you can come up with the correct answer.

I think it can be applied to any MLE as long as you get the correct Likelyhood function. Hope this helps alleviate some of the tedium involved in MLE's.

Ps i think they are eventually gonna ban this calc!

yes, thats what ive been using! that way ya dont have to factor

Hawgdriver

08-15-2010, 04:47 PM

I can't remember the last time I solved an MLE the 'long way'. For me, it's either a shortcut solution such as:

alpha = - n / K , where K = ln [(theta + d )^(n+c) / (product of x_i plus theta)]

or using the table function on the loglikelihood or likelihood function and plugging in the answers.

you laugh at me integrating using this thing. came in handy yesterday, and it's actually pretty easy to do if you practice it.

yashi911

08-15-2010, 04:58 PM

I wasn't trying to be rude, I meant crazy smart. :)

I'm good at integration by parts so your way is just too hard for me.

Hawgdriver

08-15-2010, 05:17 PM

I wasn't trying to be rude, I meant crazy smart. :)

I'm good at integration by parts so your way is just too hard for me.

No, no, I felt the love, but thanks.

Integration by parts takes me too long. Plus, my idea is that even if it's not the best tool for a given problem, it's another weapon that's available on exam day. There should be a study guide just for this calculator, really. Maybe I should write one!

No, no, I felt the love, but thanks.

Integration by parts takes me too long. Plus, my idea is that even if it's not the best tool for a given problem, it's another weapon that's available on exam day. There should be a study guide just for this calculator, really. Maybe I should write one!

Using a scientific calculator like some kind of mini-Matlab/advanced graphing calculator, then charging money for a study guide that comes out with a new edition every six months...

Sounds crazy, but you may be up to something there... :lol:

Abraham Weishaus

08-20-2010, 02:10 PM

One thing I love to do with integrals is to use a approximating technique to solve rather than say integration by part (which I always make small mistakes on). Table mode allows me to get a lot of function results without having to edit the values all the time so I can find

\int_0^{\infty}\frac{x^3e^{-x/100}}{100^3}

Using "simpson's rule (http://en.wikipedia.org/wiki/Simpson%27s_rule)":

step size = 125

Go from 0 to 1250.

\frac{125\times 2}{6}(0 + 4\times 0.560 + 2\times 1.28 + 4\times 1.24 2\times 0.842 + 4\times 0.471 + 2\times 0.233 + 4\times 0.106 +2\times 0.0454 + 4\times 0.0185 + 0.00727)=599

The rest of the integral is essentially 0:

\int_{1250}^{\infty}\frac{x^3e^{-x/100}}{100^3}=0.932

True answer = 600...

note: use the multi-line calc to get the values and a backup calculator to apply the simpon's rule

Of course you have to make sure you're spacing such that you're going through about 10% of the function dominant area each step... If your steps are too large you'll get the wrong answer... [For example a step size of 156.25 instead would get an answer of 608, step size=208.3 would give 626]

Of course the solution to the above integral should be memorized...

\frac{3!100^4}{100^3}=600

No need to memorize - it's in the tables - third moment of exponential with mean 100.

Hawgdriver

08-20-2010, 04:33 PM

No need to memorize - it's in the tables - third moment of exponential with mean 100.

and I suppose if the upper limit of integration was 500, you could use partial expectation E[(X ^ 500)^3], minus 500^3 S(500) ?

As a different (easier) approach, isn't J-Geek's problem the equivalent of finding the normalizing constant for a gamma pdf?

And I think J-Geek's approach is needlessly complicated (and it doesn't appear complicated, although I admit I don't quite follow it).

But if you don't have the required flash of intuition to solve an integral, just use a Riemann approach like me. Add up a few evenly spaced rectangles using DATA and STAT...it takes very little time.

Abraham Weishaus

08-20-2010, 04:48 PM

and I suppose if the upper limit of integration was 500, you could use partial expectation E[(X ^ 500)^3], minus 500^3 S(500) ?

Not at all: E[(X-Y)^3] does not equal E[X^3]-E[Y^3]. I like the Poisson process method for such integrals. And these are the only hard integrals that ever appear on this exam (and the Poisson process approach is easy enough), so approximate integration is not needed, and I doubt it's faster.

I agree, though, that the 30XS has other valuable features, particularly DATA combined with STATISTICS. And perhaps the TABLE can be useful for trial-and-error methods.

Hawgdriver

08-20-2010, 05:24 PM

Not at all...

Thanks, I was wondering why I couldn't get them to equate. I figured I was missing something even before I made my comment. But it is equivalent to finding the gamma normalizing constant right? To me that was more intuitive than the third moment of the exponential.

I agree with you that it's unlikely that an examinee would need to integrate something beastly for the exam. However, I am comforted that I could if I had to.

I would like to convince anyone that it is not a time-consuming endeavor. As an example of how long it takes, let me use this Riemann method using DATA and STATS for the following integral: f(x) = 2x / (x^2 -3x +4) from a = 0 to b = 4.

In L1 enter 1 thru 10

In L2 enter L1 - .5

In L3 enter L2 * .4

In L1 enter (2 L3)/(L3^2 -3L3 +4)

In STAT retrieve the sum of L1

Multiply by .4

Ok, it took 75 seconds to get 5.08, the answer is 5.07 (http://www.wolframalpha.com/input/?i=integrate+%282x+%2F+%28x^2+-3x+%2B4%29%2C+0%2C+4%29). And I haven't really done this a bunch of times like you might think. This is maybe the fourth time I've done this procedure. As you say, it seems unlikely that calculus ability will be a discriminator for the exam, but I prefer knowing that I could integrate anything numerically with a simple technique.

Edit: not familiar with using a counting process approach, is it simpler?

DovidB

09-27-2017, 12:55 PM

Here is how I used the Multi-view's data and stats functions to solve a Bayes problem.

The problem is in ASM (8th ed?) #45.26, the first bayes chapter. It was on the F00 course 4 exam, problem #28 (page 29) (http://www.soa.org/files/pdf/course4_1100.pdf).

I will show two methods for solving the problem, the first is a standard approach and the second takes advantage of the "data" function using multiview (MV).

The problem:

There are three classes of risks, each equally likely. They follow a pareto distribution with \theta=10.

1: \alpha=1

2: \alpha=2

3: \alpha=3

There is an observation of 20.

What is the probability of the next observation being greater than 30?

Method 1:

I timed this. It took me three minutes. I had already solved it twice, so this was a pretty speedy solution. I tried to go through the method like I would for an exam, but knowing the facts and method and never stalling to "think" about how to solve it. Just pure algorithm...if that makes sense.

To solve:

Make a table with three column headers, 1, 2, 3 for the values of \alpha.

put 1/3 under each.

compute the value of the probability density function

f(x)=\frac{\alpha\theta^{\alpha}}{(x+\theta)^{a+1} },\;\theta=10

for each at x=20. Put the result in the appropriate column. (.0111, .0074, .0037)

sum the results (0.0222)

divide each joint probability by the sum to find the posterior probability (.5, .333, .167)

compute the value of S(30) for each: (.25, .0625, .0156)

multiply the posteriors by the corresponding S(30) and take the sum: 0.148

Method 2, using MV:

In L1 put 1, 2, 3

In L2 put in the formula L2=L1 10^L1/30^(L1+1 {you don't need a space or "times" between L1 and 10, and you don't need a final ")" on expressions.

[2nd] [mode] {quit}

[2nd] [data] {stat}

1Var stats, L2 data, hit [5] to return \sum{x} then [sto>] [x]

[data] enter a formula into L3: L2/x

"freeze" L3 by re-entering .5 into the top cell (if this step confuses you, try to skip it and see the result)

enter a formula into L2: .25^L1

"freeze" L2 by re-entering .25 into the top cell

enter a formula into L1: L2L3 (you don't need to hit the multiply button)

[2nd] [mode] {quit}

[2nd] [data] {stat}

1Var stats, L1 data, hit [5] to return \sum{x}, which is 0.148

This took me two minutes, but I made a few mistakes and had to start over. The main time savings is due to the fact you don't have to compute 3x each of (density, posterior, S(30)). I'm convinced it's worth the effort to make the [data] function and the [stat] function your friend. Also less likelihood of error.

If method 2 seems weird, it will seem natural with a little practice.

Anyone disagree? Dissent is welcome.

Your technique for this is amazin! One quick question, in the following line: "enter a formula into L2: .25^L1", where did you get the .25 from and what is it doing?

vBulletin® v3.7.6, Copyright ©2000-2018, Jelsoft Enterprises Ltd.