README.md

In elizabethastroud/Cropprocessing: Data organisation, classification and visualisation of archaeobotanical data to understand crop processing stage

CropPro

This repository stores the CropPro package. CropPro contains functions which can be used to classify archaeobotanical data against a linear discriminant model derived from ethnographic data on the by-products of crop processing stages. The package contains two linear discriminant analysis (LDA) functions that can be used to classify the archaeobotanical samples as one of four crop processing stages, or one of five classes (four crop processing stages and the archaeological group) and then plot them as two-dimensional or three-dimensional graphs. This package also contains a function for the data transformation required before LDA can be conducted, as well as a function for plotting the ratio of grain to chaff to weed seeds in comparison to ethnographic data.

Referencing

The CropPro package draws on data from ethnographic research conducted on the Greek island of Amorgos. The models were developed by Jones (1984, 1987, 1990) and Charles (1997).

When publishing results obtained from the use of the CropPro package please cite the package and its version, as well as the ethnographic data. A best practice example paragraph is provided in Stroud et al. (2024).

Package citation: Stroud, E., Jones, G., Charles, M., Bogaard, A. (2023) CropPro: Data organisation, classification and visualistion of archaeobotanical data to understand crop processing stage. R package version 1.0.0, https://github.com/CropPro-package/CropPro

Data sources:

Crop processing LDA model:

Jones, G., (1984) Interpretation of archaeological plant remains: Ethnographic models from Greece. In (eds) W. van Zeist & W.A. Casparie, Plants and Ancient Man: Studies in Palaeoethnobotany. Rotterdam: Balkema, pp 43-61

Jones, G., (1987). A statistical approach to the archaeological identification of crop processing. Journal of Archaeological Science, 14(3), pp.311-323.

Triplot model:

Jones, G. (1990). The application of present-day cereal processing studies to charred archaeobotanical remains. Circaea 6(2):91-96

LDA plus model:

Charles, M., (1998). Fodder from dung: the recognition and interpretation of dung-derived plant material from archaeological sites, Environmental Archaeology, 1:1, 111-122

Example paper:

Stroud, E., Jones, G., Charles, M. Bogaard, A. (2024) Sieving the weeds from the grains: an R based package for classifying archaeobotanical samples of cereals and pulses according to crop processing stages. Vegetation History and Archaeobotany. https://doi.org/10.1007/s00334-024-01006-7

Installation

You can install the development version of CropPro from GitHub with:

# install.packages("devtools")
devtools::install_github("CropPro-package/CropPro")

Crop processing LDA example

This is a basic example which shows you how to use the package to understand crop processing stage represented within an archaeobotanical assemblage :

library(CropPro)
#> Loading required package: dplyr
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
#> Loading required package: MASS
#> 
#> Attaching package: 'MASS'
#> The following object is masked from 'package:dplyr':
#> 
#>     select
#> Loading required package: rgl
## basic example code
# Sample data
species<-c("Chenopodium album" , "Anthemis cotula", "Brassica rapa ssp campestris",
"Raphanus raphanistrum", "Agrostemma githago" , "Poa annua" )
category<-c("SFH", "BHH","SFH","BHH","BFH", "SFL" )
s.1246<-c(2,2,3,7,1, 6)
s.46178<-c(1,1,0,0,2,5)
s.1<-c(0,2,2,8,1,8)
s.23<-c(0,3,2,3,0,0)
s.987<-c(3,9,6,8,0,1 )
dataset<-data.frame(species,category,s.1246,s.46178,s.1,s.23,s.987)
dataset
#>                        species category s.1246 s.46178 s.1 s.23 s.987
#> 1            Chenopodium album      SFH      2       1   0    0     3
#> 2              Anthemis cotula      BHH      2       1   2    3     9
#> 3 Brassica rapa ssp campestris      SFH      3       0   2    2     6
#> 4        Raphanus raphanistrum      BHH      7       0   8    3     8
#> 5           Agrostemma githago      BFH      1       2   1    0     0
#> 6                    Poa annua      SFL      6       5   8    0     1

The sample data, showing species, classification and 5 different samples : s.1246, s.46178, s.1, s.23, s.987, needs to be transformed and transposed for the LDA. The function crop.dataorg can do that.

data<-crop.dataorg(dataset, codes=2, samples=3)
data
#>   samples       BHH      BFH SHH SHL      SFH      SFL
#> 1  s.1246  8.859570 2.182179   0   0 6.865712 5.345225
#> 2 s.46178  3.333333 4.714045   0   0 3.333333 7.453560
#> 3     s.1  9.258201 2.182179   0   0 3.086067 6.172134
#> 4    s.23 12.247449 0.000000   0   0 5.000000 0.000000
#> 5   s.987 11.216813 0.000000   0   0 8.047379 1.924501

The organised data can then be analysed using LDAcrop.pro which will classify the data against the ethnographic model.

results<-LDAcrop.pro(data)
#> [1] "Classification results and linear discriminant scores "
#>   samples Class_std* Prob.1_std* Prob.2_std* Prob.3_std* Prob.4_std*  LD1*
#> 1  s.1246          1       1.000       0.000           0           0 1.279
#> 2 s.46178          1       1.000       0.000           0           0 0.429
#> 3     s.1          1       1.000       0.000           0           0 1.744
#> 4    s.23          1       0.988       0.012           0           0 1.744
#> 5   s.987          1       0.999       0.000           0           0 1.570
#>    LD2*   LD3*
#> 1 0.315 -6.143
#> 2 0.558 -6.764
#> 3 1.143 -6.720
#> 4 0.743 -2.291
#> 5 0.045 -3.811
#> [1] "classification table"
#>                         Count Percentage
#> Winnowing by-product        5        100
#> Coarse sieve by-product     0          0
#> Fine sieve by-product       0          0
#> Fine sieving product        0          0

The results from LDAcrop.pro can be plotted as either a static 2D or interactive 3D graph.

crop.plot2D(results)

crop.plot3D(results)

LDA.plus example

There are alternative functions within the package which allow for classification of the entered data against a model constructed from the ethnographic data and the entered data. This provides the LDA an alternative group of “archaeological” when classifying the entered data. The workflow is similar to above, except with the use of the “dung” suite of functions: LDAcrop.plus, crop.plus_plot2D, crop.plus_plot3D.

plus_results<-LDAcrop.plus(data)
#> [1] "Classification results and linear discriminant scores"
#>    Sample CLASS_std* Prob.1_std* Prob.2_std* Prob.3_std* Prob.4_std*
#> 1  s.1246          5           0           0           0           0
#> 2 s.46178          5           0           0           0           0
#> 3     s.1          5           0           0           0           0
#> 4    s.23          5           0           0           0           0
#> 5   s.987          5           0           0           0           0
#>   Prob.5_std*  LD1*   LD2*   LD3*   LD4*
#> 1           1 1.935 -5.015 -6.926  0.098
#> 2           1 1.036 -4.123 -5.220 -2.164
#> 3           1 2.474 -6.041 -6.972  0.013
#> 4           1 2.184 -4.773 -5.706  3.663
#> 5           1 2.092 -4.508 -6.602  2.344
#> [1] "classification table"
#>                          Count Percentage
#> Winnowing by-product         0          0
#> Coarse sieveg by-product     0          0
#> Fine sieve by-product        0          0
#> Fine sieve product           0          0
#> Archaeological               5        100
crop.plus_plot2D(plus_results)

crop.plus_plot3D(plus_results)

elizabethastroud/Cropprocessing documentation built on Sept. 27, 2024, 3:03 p.m.