[R] approxfun returning value higher than I would expect

Thu Sep 16 12:30:19 CEST 2010

stephen sefick wrote:
> I will get R-dev tomorrow, and give it a try.  Where do I check out the svn?
>   

https://svn.r-project.org/R/trunk

The fixed files are in .../src/library/stats/R and are approx.R, 
spline.R and splinefun.R.  The only one that affects your report is 
approx.R.

Duncan Murdoch

> thanks,
>
> Stephen
>
> On Wed, Sep 15, 2010 at 10:00 PM, Duncan Murdoch
> <murdoch.duncan at gmail.com> wrote:
>   
>> I think this is fixed now.  There were actually two bugs:
>>
>> I fixed an old one a few days ago, but my fix didn't handle the case of
>> unsorted x properly.  (I haven't checked whether the old code handled that
>> properly; I'd guess not, but it might have.)  Now I've fixed my new bug.
>>
>> *Please* test R-devel to see if I got it right this time.  (I think I did,
>> but I thought I did last time, too.)  If you don't build your own copies,
>> look for a build based on revision r52916 or newer.  Windows builds should
>> appear late tomorrow, I don't know about other platforms.
>>
>> Duncan Murdoch
>>
>> On 15/09/2010 4:32 PM, stephen sefick wrote:
>>     
>>> x <-
>>> c(-0.48,-0.48,-0.42,-0.26,0.58,0.48,0.47,0.54,0.5,0.52,0.52,0.56,0.58,0.61,0.68)
>>> y <- c(0,0.2,0.4,0.6,0.8,1,1.2,1.4,1.6,1.8,2,2.2,2.4,2.6,2.8)
>>>
>>> s <- approxfun(x[4:5], y[4:5], ties=mean)
>>>
>>> s(0)
>>>
>>> #This is the value that I want.  The first zero crossing in the order
>>> of y.  #In other words in between 0.6 and 0.8.
>>>
>>> #The data is a channel cross-section, and should really be visualized
>>>
>>> plot(x~y, ylim=c(1,-1)) #this is half of a cross-section
>>>
>>> #The bankfull depth is where x is equal to 0
>>>
>>> On Wed, Sep 15, 2010 at 3:12 PM, stephen sefick <ssefick at gmail.com> wrote:
>>>       
>>>> Thanks and I'll strip the code down even more in future posts.
>>>>
>>>> Stephen
>>>>
>>>> On Wed, Sep 15, 2010 at 3:05 PM, Duncan Murdoch
>>>> <murdoch.duncan at gmail.com> wrote:
>>>>         
>>>>>  On 15/09/2010 3:48 PM, stephen sefick wrote:
>>>>>           
>>>>>> Below is the code that I am using in a much larger function.  I would
>>>>>> expect a bankfull measure at zero to be between 0.6 and 0.8 approxfun
>>>>>> is returning 0.8136986.  I am sure that I am missing something.
>>>>>>
>>>>>> measure_bkf<- (structure(list(measurment_num = c(0, 0.2, 0.4, 0.6,
>>>>>> 0.8, 1, 1.2,
>>>>>> 1.4, 1.6, 1.8, 2, 2.2, 2.4, 2.6, 2.8, 3, 3.2, 3.4), bankfull_depths_m
>>>>>> = c(-0.48,
>>>>>> -0.48, -0.42, -0.26, 0.58, 0.48, 0.47, 0.54, 0.5, 0.52, 0.52,
>>>>>> 0.56, 0.58, 0.61, 0.68, 0.62, 0.67, 0.66)), .Names =
>>>>>> c("measurment_num",
>>>>>> "bankfull_depths_m"), row.names = c("6124", "612", "613", "614",
>>>>>> "615", "616", "617", "618", "619", "620", "621", "622", "623",
>>>>>> "624", "625", "626", "627", "628"), class = "data.frame"))
>>>>>>
>>>>>>
>>>>>> measure_bkf_not_zero<- subset(measure_bkf,
>>>>>> measure_bkf$bankfull_depths_m!=0)
>>>>>>
>>>>>> bkf_min<- which.max(measure_bkf_not_zero[,"bankfull_depths_m"]<0)
>>>>>>
>>>>>> bkf_max<- which.max(measure_bkf_not_zero[,"bankfull_depths_m"])
>>>>>>
>>>>>> #bkf_min<- ifelse(length(bkf_min)>1, bkf_min[1], bkf_min)
>>>>>> #bkf_max<- ifelse(length(bkf_max)>1, bkf_max[1], bkf_max)
>>>>>>
>>>>>> #s<- with(measure_bkf_not_zero, approx(measurment_num,
>>>>>> bankfull_depths_m,
>>>>>> xout=seq(measure_bkf_not_zero[bkf_min,"measurment_num"],
>>>>>> measure_bkf_not_zero[bkf_max,"measurment_num"], length=2000)))
>>>>>> #int_bkf<- with(s, x[which.min(y[y>0])])
>>>>>>
>>>>>> s<- with(measure_bkf_not_zero[bkf_min:bkf_max,],
>>>>>> approxfun(bankfull_depths_m, measurment_num), ties=mean)
>>>>>>
>>>>>> int_bkf<- s(0)
>>>>>>
>>>>>>             
>>>>> It is easier to see the problem if you don't leave all the complications
>>>>> in
>>>>> the beginning.  Just define some variables and
>>>>> show the interpolation on a plot:
>>>>>
>>>>> x <-
>>>>>
>>>>> c(-0.48,-0.48,-0.42,-0.26,0.58,0.48,0.47,0.54,0.5,0.52,0.52,0.56,0.58,0.61,0.68)
>>>>> y <- c(0,0.2,0.4,0.6,0.8,1,1.2,1.4,1.6,1.8,2,2.2,2.4,2.6,2.8)
>>>>> plot(x, y)
>>>>> s <- approxfun(x, y, ties=mean)
>>>>> curve(s, add=TRUE)
>>>>>
>>>>> On my system, this looks okay in 2.11.1, but not in R-patched or R-devel
>>>>> (soon to be 2.12.0).  It is fixed if the x values are ordered, but it's
>>>>> not
>>>>> supposed to need that.  I'll take a look.
>>>>>
>>>>> Duncan Murdoch
>>>>>
>>>>>           
>>>> --
>>>> Stephen Sefick
>>>> ____________________________________
>>>> | Auburn University                                   |
>>>> | Department of Biological Sciences           |
>>>> | 331 Funchess Hall                                  |
>>>> | Auburn, Alabama                                   |
>>>> | 36849                                                    |
>>>> |___________________________________|
>>>> | sas0025 at auburn.edu                             |
>>>> | http://www.auburn.edu/~sas0025             |
>>>> |___________________________________|
>>>>
>>>> Let's not spend our time and resources thinking about things that are
>>>> so little or so large that all they really do for us is puff us up and
>>>> make us feel like gods.  We are mammals, and have not exhausted the
>>>> annoying little problems of being mammals.
>>>>
>>>>                                -K. Mullis
>>>>
>>>> "A big computer, a complex algorithm and a long time does not equal
>>>> science."
>>>>
>>>>                              -Robert Gentleman
>>>>
>>>>         
>>>
>>>       
>>     
>
>
>
>