[R] normalised curve fitting with error bars

[David Winsemius]

On Jul 3, 2009, at 12:52 PM, Pooja Jain wrote:

> Hi David,
>
> Sorry for not being absolutely clear.
>
> x-axis will be the values ranging from (-50 to 50) and y-axis will  
> be the density distribution as you would get after plotting every  
> column fitted to the normal distribution. However, what I need here  
> is to reduce the nine column file to a three column file. In the  
> three column file, the first column should be the mean of every row  
> of a the nine column file, the second column will be the maximum  
> value of every row of the nine column file and the third column will  
> be the minimum value of all of the values in every row of the nine  
> column file.
>

The three vectors you desire can be created with:
apply(x, 1, mean)
apply(x, 1 ,max)
apply(x ,1, min)

  In one line it could be

 > new3 <- data.frame(mean50 =apply(x, 1, mean), max50=apply(x, 1,  
max), min50= apply(x, 1, min))
Then plot density(new3$mn) and the others.

plot(density(new3$mean50), xlim=c(min(new3$min50),max(new3$max50) ) )
  points(density(new3$min50))  # may need to set the ylims if this  
overshoots
  points(density(new3$max50))

I think....

-- 
David

> Since the values in the nine column fits quite well to the normal  
> distribution, I expect their mean to behave similar. So I would plot  
> mean fitted to the normal distribution and would like to see for  
> every mean value what is the error bar as defined by the min and max  
> values.
>
> Thank you for any help that you may extend in this matter.
>
> Best regards,
> -Pooja
>
> On 3 Jul 2009, at 17:32, David Winsemius wrote:
>
>>
>> On Jul 3, 2009, at 11:56 AM, Pooja Jain wrote:
>>
>>> Dear List,
>>>
>>>
>>> My data consist of nine columns and about 50,000 rows. It looks  
>>> like this.
>>>
>>> -9.0225	3.46464	2.80926	-0.3847	3.73735	1.1058	-2.98936	 
>>> 1.38901	-8.1846
>>> -2.4315	-5.1189	1.8225	3.3798	1.7874	4.693	-3.9286	1.4266	5.7849
>>> -3.4894	-4.0305	3.7879	3.5195	2.9186	2.8685	-6.126	4.978	4.9381
>>> 4.5282	3.62558	-3.0455	4.6518	1.39746	0.68652	3.5708	-3.6404	-4.2963
>>> -1.3183	0.6752	-4.0382	-2.5386	-0.6459	1.0689	-0.6392	-6.4141	-4.101
>>> -1.3735	3.1098	-2.8291	-5.2548	-3.3798	1.3959	-1.8605	0.1522	2.1818
>>> -1.0488	0.1071	-3.7154	-1.3748	-5.6218	-0.9989	-2.6763	-4.6548	 
>>> 0.5449
>>> 7.948	2.0673	3.8729	8.0537	2.79	-3.6963	8.4584	-1.5122	-6.3354
>>> 4.5827	6.1787	-3.1787	1.4554	5.6973	5.3386	10.6077	-0.0424	3.3653
>>> 4.1653	8.0266	1.9509	4.2077	5.4182	4.1797	7.9248	1.1502	0.753
>>> 3.6046	3.6743	11.8299	9.5704	10.7384	8.675	4.9277	13.6898	13.3279
>>> 18.4431	12.3946	22.1126	22.7109	19.4623	17.3565	15.27	17.5922	 
>>> 19.5873
>>>
>>>
>>> The values in each of the nine columns are almost normally  
>>> distributed but have slightly different heights of nine normalised  
>>> bell shaped plots.
>>>
>>> I was trying to plot each row as a point with error bars. The  
>>> central point will be the mean of the nine values in each of rows  
>>> and the max and min of the row will be the upper and lower bound  
>>> of the error bar.  I expect a normally distributed plot with  
>>> spikes around. Can any one please help me plotting data in this  
>>> way ?
>>>
>>> Many thanks for any input.
>>>
>>
>> I am confused about what is desired. First you describe the  
>> distributions of the full dataset's columns. Then you ask for a  
>> plot by rows, but you say it will be a normally distributed plot  
>> (which it is not). What are the x-axis and y-axis supposed to be?
>>
>> David Winsemius, MD
>> Heritage Laboratories
>> West Hartford, CT
>>
>
>
> This message has been checked for viruses but the contents of an  
> attachment
> may still contain software viruses, which could damage your computer  
> system:
> you are advised to perform your own checks. Email communications  
> with the
> University of Nottingham may be monitored as permitted by UK  
> legislation.
>

David Winsemius, MD
Heritage Laboratories
West Hartford, CT