[R] How to manipulate tables

[David Winsemius]

On Dec 26, 2009, at 4:56 PM, James Rome wrote:

> Thanks David.
>
> I wanted to calculate the Poisson distribution from my histograms to  
> see
> how closely they match it. So the formula is something like
> pprob=((lambda**cnts)/factorial(cnts))*exp(lambda)

(Please don't reply privately.)

You have two vectors of different lengths, the lambdas and the counts,  
and you need to create the 2-way combinations in a form that can be  
supplied to a function.

> And I was hoping to do it for the whole table at once.

Something like this?

 > lambdas <-scan()   # scan can read a sequence of numbers separated  
by white space.

1: 0.199190283 0.013765182 0.006477733 0.017813765 0.093117409  
0.160323887
7: 0.401619433
8:
Read 7 items

 > ldf <- expand.grid(lambdas, 1:15)
 > str(ldf)
'data.frame':	105 obs. of  2 variables:
$ Var1: num  0.19919 0.01377 0.00648 0.01781 0.09312 ...
$ Var2: int  1 1 1 1 1 1 1 2 2 2 ...
- attr(*, "out.attrs")=List of 2
  ..$ dim     : int  7 15
  ..$ dimnames:List of 2
  .. ..$ Var1: chr  "Var1=0.199190283" "Var1=0.013765182"  
"Var1=0.006477733" "Var1=0.017813765" ...
  .. ..$ Var2: chr  "Var2= 1" "Var2= 2" "Var2= 3" "Var2= 4" ...

 > ldf$countprob <- (ldf$Var1**ldf$Var2)/ factorial(ldf$Var2)*exp(ldf 
$Var1)
 > ldf
           Var1 Var2     countprob
1   0.199190283    1 2.430946e-01
2   0.013765182    1 1.395597e-02
3   0.006477733    1 6.519830e-03
4   0.017813765    1 1.813394e-02
snipped remaining 100 lines...

-- 
David

> Jim Rome
>
> On 12/26/09 2:28 PM, David Winsemius wrote:
>>
>> On Dec 26, 2009, at 2:02 PM, James Rome wrote:
>>
>>> I am sorry to be bothering the list so much.
>>>
>>> I made a table of counts of flight arrivals by hour:
>>
>> No, you made a list of tables, which is different.
>>
>>> cnts=tapply(Arrival4,list(Hour),table). There are up to 15  
>>> arrivals in a
>>> bin.
>>
>> Why not work with something more regular, like (this is all guesswork
>> and untested in absence of test data objects):
>>
>> table(Arrival4, factor(Hour, levels=1:15)  )    ?
>> # or use whatever your max(arrivals) might be.
>>
>>> xx <- factor(sample(1:10, 4))
>>> xx
>> [1] 9 4 7 1
>> Levels: 1 4 7 9
>>
>>> table(factor(xx, levels=1:10) )
>>
>> 1  2  3  4  5  6  7  8  9 10
>> 1  0  0  1  0  0  1  0  1  0
>>
>>
>>
>> cnts
>>>
>>> $`0`
>>>
>>> 1  2  3  4  5  6  7  8  9 10 13
>>> 1  2  5  9  2  7  5  4  2  4  1
>>>
>>> $`1`
>>>
>>> 1 2 3 4
>>> 3 2 2 1
>>>
>>> $`2`
>>>
>>> 1 3
>>> 2 2
>>> .  . .
>>>
>>> My first problem is how to get this table filled in with the 0  
>>> slots.
>>> E.g., I want
>>> $`0`
>>>
>>> 1  2  3  4  5  6  7  8  9 10 11 12 13 14 15
>>> 1  2  5  9  2  7  5  4  2  4   0   0   1   0   0
>>>
>>> for all 24 hours. The elements of the tables are lists, but I do not
>>> seem to be able to extract the names of each list, which I think  
>>> would
>>> allow this manipulation.
>>>
>>> My second problem is that I want to compute probability  
>>> distributions. I
>>> have
>>> lambda
>>>         0           1           2           4           5
>>> 6           7
>>> 0.199190283 0.013765182 0.006477733 0.017813765 0.093117409  
>>> 0.160323887
>>> 0.401619433
>>>         8           9          10          11          12
>>> 13          14
>>> 0.191093117 0.177327935 0.318218623 0.404858300 0.463157895  
>>> 0.495546559
>>> 0.435627530
>>>        15          16          17          18          19
>>> 20          21
>>> 0.418623482 0.307692308 0.405668016 0.484210526 0.580566802  
>>> 0.585425101
>>> 0.519028340
>>>        22          23
>>> 0.556275304 0.503643725
>>>
>>> and I need to calculate lambda**cnts for each bin, and each hour.  
>>> I am
>>> also unsure of how to do this.
>>
>> Can you give a justification for this procedure? (I'm not saying it  
>> is
>> nonsense, only that it does not make make sense to me. So perhaps
>> there is some mathematical property about which I need education.)
>>

David Winsemius, MD
Heritage Laboratories
West Hartford, CT