hanks to everyone who suggested solutions for my
last problem - fitting distribtuions to a cumulative
data set.<br><br>I have now fitted a distribtuion to
my data set and run a simulation. With the
forecasted values I've fitted a distribtuion (Weillbull) and
recorded the associated parameters (e.g. scale and shape,
setting the location to one myself).<br><br>I now have
another dataset which shows the cumulative proportion
healed at week 36 only. I want to be able to fit my
Weilbull distribution (and associated parameters) to this
single point so it can give some estimate of the
proportions healed in the weeks before and after week
36.<br><br>So far I cannot think of a way around this? Is it
possible and if so has anyone got any ideas?<br><br>All
suggestions will be gratefully received!<br><br>James

ctually, your parameters are being accepted, but
the numbers are being rounded for display. To see the
significant digits of the decimal, just change the number
formatting of the assumption cell in the spreadsheet to show
the number of digits you want.<br><br>I hope that
helps; if there are any further questions, please email
tford@....<br><br>Thank you,<br>Eric Weissmann<br>Chief Executive
Officer<br>Decisioneering, Inc.

an't find a solution for this problem in the
manual nor in help. I wish to choose a distribution
(uniform or triangular) for fractional values between 0.0
and 0.0056, but CB resets the min and max values to
0.0 and 0.1 after I enter them, not allowing my
assumptions. I tried entering the number of decimal places in
the format preferences section of the run window, but
that doesn't work either. <br><br>Would appreciate any
tips.

ne an assumption cell<br>Use excel PV function to
reference <br>each assumption cell<br><br>The cell you use
for PV function also define as<br>forecast cell.

as anyone a suggestion for this problem? I want
to compute the net present value for a problem in
which at least three of the arguments in the function
are subject to probability distributions. It is not
clear how I can use the present value routine in Excel
with CB in this context. No suggestions in the
manual.<br><br>hcgoddard

rystal Ball can not fit probability functions to
cumulative data. It can only fit to datasets containing
actual values or frequency information.<br><br>If you
have cumulative data, you should define a custom
distribution and import the cumulative data using the Data
button (see pg. 88-89 of the user manual) You should
also provide the first data point with a probability
of 0.

runcating a distribution actually changes the
shape of the distribution you are defining. So in the
example you gave, the frequency (or probability) of
values above the truncation point are scaled up by
100/84.<br>I hope this helps.

e've now realised how crystal ball interpets the
data when fitting a distribution. Instead of
interpreting the first row/column as weeks and the second as
frequency it seems to be combining them all together and
assuming that week numbers are also patients!<br><br>Is
there a way around this? The datasets we have show the
proportion healed at certain time points and are not
individual patient data. At the moment we are having to type
in one week value everytime a person heals in a
week. When you have hundreds healing in a given week
this is quite time consuming (even using the copy or
fill command)!<br><br>Can anyone help?<br><br>James
and Louise

ould you help? We're new to Crystal
Ball!<br><br>We have a dataset that measures "time to heal a
disease" in weeks. The first heal and datapoint is at week
eight. In other words nobody can heal before week 8. We
have the weekly numbers healing each week upto week
48. We now list the week number in row A and the
corresponding cumulative proportion healed in row B for each
specified week.<br><br>We now define a cell for a
distribution assumption and fit a distribution to the data. We
are entering a value of one in this cell - is this
correct?<br><br>Our concerns are:<br><br>a) Do we need to add week 1
and 0 healing at the beginning of the dataset?<br>b)
The culmulative diagram displayed in the distribution
fitting gallery does not seem to be representative of our
dataset. (e.g. in the dataset, by week 16 only 25% of
patients have healed. In the cumulative diagram it is
saying there is a correpsonding probability at week 16
of 60%)<br><br>Any advice would be very much
appreciated!<br><br>James and Louise

e wish to understand the effect of truncating a
distribution. For example, say a Normal distribution is
truncated (below) at a value equal to Mean minus Standard
Deviation, so that 16% of the values which would (in the
absence of truncation) have been generated are excluded,
does this result in the lower limit itself occurring
with frequency 16%, or in the frequency of all higher
values being scaled up by 100/84?<br>Thank you for any
responses.<br>Peter

ecause my company didn't buy the
AMI-Handbook<br>and decisioneering haven't been answered all<br>my
emails concerning delivering this documentation, now I
try this way:<br>I was very glad, to read in this
group the rigth<br>command to run CB direct from VBA
(Msg 5 and 19)and now I hope to get the command to
change the<br>ExtractData Settings direct from VBA (I'd
like to change the settings from percentile to
cumulative counts and reverse.<br>So I'm looking forward,
that somebody would post this special
command,<br><br>Read U<br>Jan!

t might seem to be a simple question but I am
new to CB-Pro.<br>I am trying to enter decesion
variables/forecast variables to a spreadsheet. It does allow me to
do that (the color of the cell changes) but when I
try to select the assumptions ot forecasts, it says
that "0 assumptions selected".<br>Now keeping inside
the same excell session, when I load another
spreadsheet and try to create these variables, I am able to
do so and run simulations.<br>So It must be some
settings in that spraedsheet that even though they allow
me to set the cells does not classify them as
decesion variables or forecast variables.<br>I checked the
documentation and it says thet either I might have used some
special char.s in the worksheet name or the sheet itself
must be corrupted.<br>Can you help me out?

ello, I am new to using CB.  Has anyone built a simple fault tree and event tree
that I could review?  I am trying to set up a risk model using CB.  Tahnks 
Barry<br>fb4060@...

re you referring to CB Predictor? If so, you can
certainly forecast with missing data. I suggest you take
the average of the data before and after the gap and
fill in the missing data with this average or you can
fill in the gap with zero's depending on your
situation.<br><br>Please feel free to contact me at Decisioneering
technical support with any comments or questions you may
have.<br><br>Regards,<br>Tracy Ford<br>Support Engineer<br>Decisioneering,
Inc.<br>helpdesk@...<br>1-800-373-5885

need to create a macro in Excel in wich i can
run the CB, but i don't now how.<br>I've tried to do
this, but when i edit the visual basic code, there are
no procedures of CB.<br>Can someone tell me how to
do that?<br>Thanks for help!<br><br>Pd: I have to
solve this problem very quickly<br>because it's
necessary for my master tesis!!!!!!!!

s there a way to use Crystal Ball Forecasting
when some of the data is missing?<br><br>For example,
suppose I have sales for the past year, but I am missing
a block of two weeks halfway through the data.

took a Decision Analysis class from Professor
Sam Savage at Stanford University. In the class, he
gave us a simple spreadsheet example of a Markov Chain
which I have used over and over again as a starting
point for my own analyses. Send me an email at my work
(jfbell@...) with "Markov Spreadsheet" as the subject and I
will send it to you.<br><br>I have always found it
easier to start with an example of a similar problem,
than to start from scratch.

s there a way to add two or more forecast
histograms to get one histogram. In addition is there a way
to weight this process. For example, I have an
object with radius uniformly distributed between -1 and
+1 and am computing its area over a number of
trials. I also have another object with uniformly
distributed radius from -2 to +2 and am running the same
number of trials. I see how CB overlays the two area
forecast histograms but I also want to be able to get a
combined distribution that I can either just add or weight
with 60% of the first object and 40% of the second?

oes anybody know anything about these and any
easy tom understand references?<br>I'm writing a model
to perform multiple simulations where some
parameters have time dependencies on the previous time
step.<br>A statistician freind of mine warned me about them
and that they *may* significantly affect quality of
result - but that's as far as he went.<br>He's not
available to help me at the moment - so anybody out there
able to?<br>Thanks <br>Bill Bailey

rystal Ball does not use Excel 4 macros any more.
Instead, it uses Visual <br>Basic interface. Therefore CB
should be called as:<br><br>CB.Simulation
Range("MAX_TRIALS").Value<br><br>You have to establish a reference to Crystal
Ball
from VBA module (as explained in AMI manual).<br>
<br>Or, at a last resort you can call it
as:<br><br>Application.Run "CB.Simulation",
Range("MAX_TRIALS").Value<br><br>Regards. Andrey.

e does have 'set to original values' in the
preferences. <br>Once I called Ken back, he believes he has
solved his problem. He has been single stepping the
first iteration and then ran the sim. I explained to
him what he found .. the single step has changes the
values and counted the changes as the original values.

i, this is Tracy at Decisioneering Tech
Support.<br>This case has been closed, but wanted to post a
solution. <br><br>It seems that Jason had long 'sheet'
and/or 'workbook' names as well as 'cell references'. We
recommend shorting your 'workbook' and 'sheet' names to 3-4
characters. Please remove any 'special' characters such as (~
& * $ #).

hen I reset the simulation, my assumption cells
do not revert back to their original values. I have
the "original values" box checked in Run Preferences.
Anyone have similar problems and/or a suggested
solution?<br><br>Thanks.

am running CB (version 4c) and am having
trouble with some of the defined assumptions - they do
not always all run! Basically I can overcome this
problem by 'kick starting' the model using the single
step function, however I am not completely sure that
this works without fail.There is nothing particularly
unusual about my model, except perhaps that all
assumption parameters are defined using external references
(which I consider to be 'best practice'). Nor am I using
any Excel add-ins (apart from the analytical toolpack
which is necessary).This is, of course, a very
dangerous phenomenon: it can significantly alter the
results of the analysis, and there is no warning that
there might be a problem. Note that it is not always
the same assumptions that do not run. Confidence can
only be ensured by a combination of looking though the
model as CB is running it to ensure that all
symmetrical assumptions are running, together with looking at
the mean values given for assumptions in the output
report for those assumptions that it would not be time
profitable to confirm that they are working (e.g. binomial
with n=1 & p<0.01)I would be greatful for any
advice or suggested reasons for this so that I could
avoid it at the design stage in further models.Jason
Petch

'm running CB v4.0c in Excel97 from a VB program
and using commands such as
Application.ExecuteExcel4Macro ("CB.Simulation(" & Range("MAX_TRIALS").Value &
")") to run the Monte Carlo simulations. But I keep
getting "No RETURN() or HALT() function found" from this
command (but not every time, and sometimes not until
several hundred trials have been done). What is going on?
Is there another way to run simulations from
VB?<br><br>Thanks<br>Marcus Fowler<br>mfowler@...

s the CEO of Crystal Ball's maker,
Decisioneering, I'm excited that there will be a user's group for
CB. I'll monitor the group from time to time. Please
feel free to send my any comments by
email.<br><br>Best regards,<br>Eric Weissmann

think this is a great idea. I will be
monitoring the chat room and help out where I can, and I
will glean ideas for enhancements from our
users.<br><br>Terry Hardy<br>Senior Quality Assurance
Engineer<br>Decisioneering, Inc.

ontacts with other users, Decisioneering software
developers, and Decisioneering support engineers.<br>*Share
best practices with other users to leverage off of
each other's experiences.<br>*Showcase individual
success with CB and gain positive recognition for CB
users and Decisioneering.<br>*Influence the future
direction of Decisioneering software.<br>*Preview new
Decisioneering and other third-party software.<br>*Attend Round
Table discussions with CB experts<br>*Hear in-depth,
how-to presentations from users and tutorials from
Decisioneering.<br><br>*There is an annual conference and a print/electronic
newsletter in the works. Keep your fingers
crossed.<br><br>Devin Tau<br>devin_tau@...<br>Business
Analyst<br>Mentor Graphics Corporation

Welcome, This is the Yahoo! Message Board for CBUG community.