In priviere/PPBstats: An R package to perform analysis found within PPB programmes

Communicate results to farmers {#com}

Within a PPB programme, it is important to give results back to the farmers in order to discuss and accompagny them in their selection. This can be done through the creation of a report with results coming from the analysis as described in section \@ref(com-technic).

The results coming from the analysis must be taken with caution by a group or farmers as it can influence collective organisation as described in section \@ref(com-rules).

Technical aspects {#com-technic}

The workflow to create the report follow three steps:

1. run the analysis and create a list with all results
2. create a .Rmd template that will be shared by all locations
3. run rmarkdown::render() in order to generate report for each location

Run the analysis

Format data

data("data_model_GxE")
data_model_GxE = format_data_PPBstats(data_model_GxE, type = "data_agro")

Descriptive plots for each location

vec_locations = levels(data_model_GxE$location)
list_hist_locations = lapply(
  vec_locations, function(x){
    p = plot(
      data_model_GxE, plot_type = "histogramm",
      vec_variables = c("y1")
      )
    p$y1
    }
  )
names(list_hist_locations) = vec_locations

Note that it is important that each element of the list refers to data of a given location in order to catch the right information in the next step when generating the report.

GGE model for all locations

The function workflow_gxe() is coming from section \@ref(ammi).

vec_variables = c("y1")
res_gge = lapply(vec_variables, workflow_gxe, "GGE")
names(res_gge) = vec_variables

Create a list with all results

res = list("hist_locations" = list_hist_locations, 
           "res_gge" = res_gge
           )

Create a .Rmd template

Your template call two objects:

• params which is list with parameter of the report
• res which is a list with all results coming from the analysis

Below is an example of a minimal template:

```{yaml, eval=FALSE}

title: "r params$title" date: "r format(Sys.time(), '%d %B %Y')" author: Flower Seed output: html_document: toc: TRUE toc_float: TRUE

---

````r

`r ''````r
knitr::opts_chunk$set(echo=FALSE, warning=FALSE, message=FALSE)

Introduction

Here is a beautiful report with all results of the year !

Description of data on the location

`r ''````r p = res$hist_locations[[location]] if(is.null(p)){print("No data for y1")}else {p}

# GGE analysis for the three locations

The germplasm effects are the following:

`r ''````r
res$res_gge$y1$p_out_mean_comparisons_gxe$germplasm

The which won where plot is the following:

`r ''````r res$res_gge$y1$p_out_biplot_gxe$biplot$which_won_where

````

### Generate a report for each location

To generate the report, you need the `R` package `rmarkdown` installed.
In the following example, the output is `.html`. 
You can choose `.pdf` or `.docx`.
See `?rmarkdown::render` for more information.

```r
library(rmarkdown)

vec_locations = names(res$hist_locations)
for (location in vec_locations){ # For each location, render a report
  params = list("title" = paste("Agronomic analyses for", location))
  rmarkdown::render("./template.Rmd", 
                    output_file =  paste("report_", location, "_", ".html", sep = ""), 
                    output_dir = "."
                    )
  }

The report generated can be visualized for loc-1, loc-2 and loc-3.

More examples of worklows with script, template and outputs can be found here.

Organisational aspects on collective work {#com-rules}

Data analysis can influense the organisation of a group. It can have advantages and disadvantages.

One advantage is that the results are discussed by all actors in the group bring information to the group and provide support for discussions between research teams and practicioners collectives

On the other hand, the analyzes are necessarily reductionist. Be careful to discuss the results with all the actors in the group to avoid analyzes disconnected to the context. Be careful also not to constrain too much or standardize the practices to meet statistical standards.

The decision of the farmer to select is based on different elements for example

• his experience on his farm
• informal exchanges with farmers, researchers, technicians, ...
• information analysed with statistics

What is the place of these elements in the decision to select a variety or a bouquet of spikes ?

It seems important to discuss these issues when sharing results among farmers, researchers and facilitators.

priviere/PPBstats documentation built on May 6, 2021, 1:20 a.m.