ptools

The library ptools is a set of helper functions I have used over time to help with analyzing count data, e.g. crime counts per month.

Installation

Hopefully in the future this will be on CRAN, but in the meantime, you can install this via devtools:

library(devtools)
install_github("apwheele/ptools", build_vignettes = TRUE)
library(ptools) # Hopefully works!

Examples

Here is checking the difference in two Poisson means using an e-test:

library(ptools)
e_test(6,2)
#> [1] 0.1748748

Here is the Wheeler & Ratcliffe WDD test (see help(wdd) for academic references):

wdd(c(20,20),c(20,10))
#> 
#>  The local WDD estimate is -10 (8.4)
#>  The displacement WDD estimate is 0 (0)
#>  The total WDD estimate is -10 (8.4)
#>  The 90% confidence interval is -23.8 to 3.8
#>    Est_Local     SE_Local Est_Displace  SE_Displace    Est_Total     SE_Total 
#>   -10.000000     8.366600     0.000000     0.000000   -10.000000     8.366600 
#>            Z        LowCI       HighCI 
#>    -1.195229   -23.761833     3.761833

Here is a quick example applying a small sample Benford’s analysis:

# Null probs for Benfords law
f <- 1:9
p_fd <- log10(1 + (1/f)) #first digit probabilities
# Example 12 purchases on my credit card
purch <- c( 72.00,
           328.36,
            11.57,
            90.80,
            21.47,
             7.31,
             9.99,
             2.78,
            10.17,
             2.96,
            27.92,
            14.49)
#artificial numbers, 72.00 is parking at DFW, 9.99 is Netflix
fdP <- substr(format(purch,trim=TRUE),1,1)
totP <- table(factor(fdP, levels=paste(f)))
resG_P <- small_samptest(d=totP,p=p_fd,type="G")
print(resG_P) # I have a nice print function
#> 
#>  Small Sample Test Object 
#>  Test Type is G 
#>  Statistic is: 12.5740089945434 
#>  p-value is:  0.1469451  
#>  Data are:  3 4 1 0 0 0 2 0 2 
#>  Null probabilities are:  0.3 0.18 0.12 0.097 0.079 0.067 0.058 0.051 0.046 
#>  Total permutations are:  125970

Here is an example checking the Poisson fit for a set of data:

x <- rpois(1000,0.5)
check_pois(x,0,max(x),mean(x))
#> 
#>  mean: 0.528 variance: 0.52373973973974
#>   Int Freq       PoisF     ResidF Prop       PoisD      ResidD
#> 1   0  586 589.7833576 -3.7833576 58.6 58.97833576 -0.37833576
#> 2   1  319 311.4056128  7.5943872 31.9 31.14056128  0.75943872
#> 3   2   79  82.2110818 -3.2110818  7.9  8.22110818 -0.32110818
#> 4   3   14  14.4691504 -0.4691504  1.4  1.44691504 -0.04691504
#> 5   4    1   1.9099279 -0.9099279  0.1  0.19099279 -0.09099279
#> 6   5    1   0.2016884  0.7983116  0.1  0.02016884  0.07983116

Here is an example extracting out near repeat strings (this is improved version from an old blog post using kdtrees):

# Not quite 15k rows for burglaries from motor vehicles
bmv <- read.csv('https://dl.dropbox.com/s/bpfd3l4ueyhvp7z/TheftFromMV.csv?dl=0')
print(Sys.time()) 
#> [1] "2022-12-15 15:31:19 EST"
BigStrings <- near_strings2(dat=bmv,id='incidentnu',x='xcoordinat',
                            y='ycoordinat',tim='DateInt',DistThresh=1000,TimeThresh=3)
print(Sys.time()) #very fast, only a few seconds on my machine
#> [1] "2022-12-15 15:31:21 EST"
print(head(BigStrings))
#>             CompId CompNum
#> 000036-2015      1       1
#> 000113-2015      2       1
#> 000192-2015      3       1
#> 000251-2015      4       1
#> 000360-2015      5       1
#> 000367-2015      6       1

Contributing

Always feel free to contribute either directly on Github, or email me with thoughts/suggestions. For citations for functions used, feel free to cite the original papers I reference in the functions instead of the package directly.

Things on the todo list: