Menu Design Proposition for Agriculture Research Design and Data Analysis

[fran2or]

With reference to #6620, the special agriculture research menu in the current version in R-Instat is Options by Context but needs to change. The new proposed name change is Experimental . The menu has not been fully conceptualized but the main proposed frame work for its menu design is as follows:

Experimental

• Open From Library
• Field Book Design
• Check Data
  • One Variable summarise
  • One Variable Graph
  • One Variable Frequencies Tables
  • Duplicates
  • Compare Columns
  • Pivot Tables
• Prepare
  • Column Operations
    • Row Summaries (Mean , Sum, Diference, etc)
    • Combine Columns
    • Recoding Columns
  • Calculate Difference Between Options
  • Stack
  • Unstack
  • Merge Additional Data
• Define
  • On - Station
  • On - Farm
• Describe
  • Frequency Tables
  • Summary Tables
  • Compare Two Options
  • Boxplot
  • Density Plot & Histograms
  • Heat maps
• Model
  • Fit Model
  • Model Comparison

Attached is slightly more detailed document. Fully prepared and approved dialogs will follow when ready.
Experimental Data Design.docx

[derekagorhom]

@fran2or from the first R script which is the Dataframe for Treatment;
the code runs well in R studio but when used in R-instat, we need to change some parts of the code so as to produce a dataframe in R-instat.
Below is the original rcode:

# Data Frame
# Multiple treatement
library("tidyr")
vec1 <- c("A", "B", "AA", "CDE")    # Create example vectors
vec2 <- 1:3
crossing(vec1, vec2) 

# Simple treatement
`data <- data.frame(data=matrix(data=NA, nrow=10, ncol=1))`

To make it run in R-instat this code should work

# Create example vectors
vec1 <- c("A", "B", "AA", "CDE")    
vec2 <- 1:3

# Create a crossing of vec1 and vec2
crossed_data <- crossing(treat = vec1, yield = vec2)

# Import the data using data_book$import_data
data_book$import_data(data_tables = list(data3 = crossed_data))

# Remove the crossed_data object from the environment
rm(crossed_data)

But even in this case that is not how the R-instat code produces new data frames:
the example below shows how we create a dataframe in R-instat.

data <- data.frame(....Code....)
data_book$import_data(data_tables=list(data=data))
rm(data)

I will try and have a meeting with lily by Monday and discuss how we can make the rcode better.

[Emmanuel-Afrifa]

I just run the codes produced above in R-INSTAT and it produces a result.

[lilyclements]

@derekagorhom thanks for explaining! So is it going to be as a new tab at the top of the "New Data Frame" dialog?

[derekagorhom]

@lilyclements yes please that is the plan since we currently do not have a way to create a dataframe for Agricultural research in R-instat

[lilyclements]

@derekagorhom Nice. Have you planned out the design of it?

[derekagorhom]

Data for treatments.odt
Yes @lilyclements above document contain the initail design.
you can have a look

[IssoufKon]

### Field book design

library(agricolae)

**##Data : trt and trt1**
trt=c("CIP-101","CIP-201","CIP-301","CIP-401","CIP-501","CIP-601")
trt1=c("a","b","c","d","e")
r=3
kinds = "Super-Duper"
serie=2
#Bloc size
k=3

[IssoufKon]

1. Completely randomized design

#Use: design.crd(trt, r, serie = 2, seed = 0, kinds = "Super-Duper",randomization=TRUE)
outdesign <-design.crd(trt1,r,serie,2543,"Mersenne-Twister")
outdesign

# no seed
outdesign <-design.crd(trt,r,serie=3)
outdesign 

2.Randomized complete block design

#Use: design.rcbd(trt, r, serie = 2, seed = 0, kinds = "Super-Duper", first=TRUE,continue=FALSE,randomization=TRUE )
trt=c("F1V1","F1V2","F1V3","F2V1","F2V2","F2V3")
outdesign <-design.rcbd(trt,r=2,serie=2,986,"Wichmann-Hill") # seed = 986
outdesign
#book <-outdesign$book # field book
fieldbook <- zigzag(outdesign)
fieldbook
#print(outdesign$sketch)
#print(matrix(fieldbook[,1],byrow=TRUE,ncol=5))

3. Latin square design

#Use: design.graeco(trt1, trt2, serie = 2, seed = 0, kinds = "Super-Duper",randomization=TRUE)
outdesign <-design.lsd(trt,serie, seed=23)
outdesign
# Write on hard disk.
# write.table(lsd,"lsd.txt", row.names=FALSE, sep="\t")
# file.show("lsd.txt")

4. Graeco-Latin designs

#Use: design.graeco(trt1, trt2, serie = 2, seed = 0, kinds = "Super-Duper",randomization=TRUE)
outdesign <- design.graeco(trt,trt2,serie=1)
outdesign 

5. Incomplete Latin Square Design or youden Design

#Use: design.youden(trt, r, serie = 2, seed = 0, kinds = "Super-Duper",first=TRUE, randomization=TRUE)
outdesign <-design.youden(trt,r,serie,seed=23)
outdesign 
#youden <- outdesign$book
#print(outdesign$sketch)
#plots <-as.numeric(youden[,1])
#print(matrix(plots,byrow=TRUE,ncol=r))
#print(youden) # field book.
# Write on hard disk.
# write.table(youden,"youden.txt", row.names=FALSE, sep="\t")
# file.show("youden.txt")

6. Balanced Incomplete Block Designs

# Use: design.bib(trt, k, r=NULL, serie = 2, seed = 0, kinds = "Super-Duper",maxRep=20,randomization=TRUE)
outdesign<-design.bib(trt,r=5, k, serie, seed =41, kinds="Super-Duper") # seed = 41
#print(outdesign$parameters)
#book<-outdesign$book
#plots <-as.numeric(book[,1])
#matrix(plots,byrow=TRUE,ncol=k)
#print(outdesign$sketch)
# write in hard disk
# write.csv(book,"book.csv", row.names=FALSE)
# file.show("book.csv")

7. Cyclic designs

#Use: design.cyclic(trt, k, r, serie = 2, rowcol = FALSE, seed = 0, kinds = "Super-Duper",randomization=TRUE
trt<-letters[1:8]
# block size = 2, replication = 6
outdesign1 <- design.cyclic(trt, k=3, r=3, serie)
outdesign1 
#names(outdesign1)
 # groups 1,2,3
#outdesign1$sketch[[1]]
#outdesign1$sketch[[2]]
#outdesign1$sketch[[3]]
#outdesign1$book
 # row-column design
#outdesign2<- design.cyclic(trt, k=2, r=6, serie=2, rowcol=TRUE)
#outdesign2$sketch

8. Alpha designs

# use : design.alpha(trt, k, r, serie = 2, seed = 0, kinds = "Super-Duper",randomization=TRUE)
trt<-1:30
outdesign<- design.alpha(trt,k,r,serie)  
outdesign

9. Augmented block designs

#Use:design.dau(trt1, trt2, r, serie = 2, seed = 0, kinds = "Super-Duper", name="trt",randomization=TRUE)
trt<-c("A","B","C","D")
#trt1<-letters[20:26]
outdesign <-design.dau(trt,trt1, r, serie)
outdesign

10. Split plot designs

# Use: design.split(trt1, trt2,r=NULL, design=c("rcbd","crd","lsd"),serie = 2, seed = 0, kinds = "Super-Duper", first=TRUE,randomization=TRUE)
# 4 treatments and 5 blocks in split-plot
trt<-c("A","B","C","D") # have to be a character
trt1<-c(1,2,3) 
outdesign <-design.split(trt,trt1,r=3,serie=2,seed=45,kinds ="Super-Duper")#seed=45
outdesign 
# book<-outdesign$book# field book
# write in hard disk
# write.table(book,"book.txt", row.names=FALSE, sep="\t")
# file.show("book.txt")

11. Split Plot Design

#Use: design.split(trt1, trt2,r=NULL, design=c("rcbd","crd","lsd"),serie = 2,seed = 0, kinds = "Super-Duper", first=TRUE,randomization=TRUE)
outdesign <-design.split(trt,trt1, r=3,serie=2,seed=45,kinds ="Super-Duper")#seed=45
outdesign
book<-outdesign$book# field book
# write in hard disk
# write.table(book,"book.txt", row.names=FALSE, sep="\t")
# file.show("book.txt")
design.strip 

12. Strip Plot Design

#Use:design.strip(trt1, trt2,r, serie = 2, seed = 0, kinds = "Super-Duper",randomization=TRUE)
trt=c("A","B","C","D","E","F")
outdesign <-design.strip(trt,trt1,r, serie=2,seed=45,kinds ="Super-Duper") # seed = 45
#book <-outdesign$book # field book
# write in hard disk
# write.table(book,"book.txt", row.names=FALSE, sep="\t")
# file.show("book.txt")

13. Design of experiments for a factorial

#Use: design.ab(trt, r, serie = 2, design=c("rcbd","crd","lsd"), seed = 0, kinds = "Super-Duper",first=TRUE,randomization=TRUE)
# factorial 3 x 2 with 3 blocks
trt<-c(3,2) # factorial 3x2
outdesign <-design.ab(trt, r=3, serie=2)
outdesign 
book<-outdesign$book
head(book,10) # print of the field book

[lilyclements]

This looks really interesting @IssoufKon. Are these suggestions for a dialog which will generate the data for this?

How do you propose we offer these options? I can't remember the differences between, say, a completely randomised design and a randomised complete block design, but, perhaps it would be good for the user to say what they want, and for us to offer the appropriate design as a result (so it's guided by analysis not by definition). Similar to how our modelling dialogs would work - the user would put in their variables, and we offer the appropriate model for it. Or similar to how we do our One Var Graph/Summarise where the appropriate graphs and statistics are offered.

On some tricot analyses we are doing, I've come across a function randomize which is an "incomplete block A-optional design". I'm not too sure on how this differs from an incomplete block design, etc, but thought I'd run it by you here in case either you knew, and/or if it were already covered in your 13 options.

[IssoufKon]

Hi @lilyclements,
We have used the design types in the agricolae package. The difference between “Completely Randomized Design” and “Randomized Complete Block Design” is that “Completely Randomized Design” generates a completely randomized design with equal or different repetitions, whereas “Randomized Complete Block Design” generates a completely randomized block design. We'd like the user to fill in the various boxes, and based on this, only those design types corresponding to these criteria will appear. The dialog is intended to produce design data that can then be used for graphic visualization.

[lilyclements]

Great, so how can we put that on a dialog as a different option? Perhaps if they fill out the "Block" receiver it gives a RCBD and otherwise gives a CRD?

Let's continue on this! How does a Latin Square differ to a Greco-Latin Square?

[derekagorhom]

#9459
Hello guys the above link goes to the first feature for agricultural research design (the plot option dialogue)