Item Based Collaborative Filterign For Multi-Trait and Multi-Environment Data in R - Development version 1.2 - rev 5.

[Last README update: 2018-03-03]

Table of contents

• NEWS
• instructions
  • Installation
  • Load the package
  • Example of Cross-validation with IBCF.MTME and external data
    • Load external data
    • Generate a data set in tidy data
    • Generate a Cross-validation
    • Fitting the predictive model
    • Show some results
  • Example of Years prediction with IBCF.Years Function
    • Loading your data
    • Transforming the data from Tidy data to matrix form
    • Adjust the model
    • Show some results
  • Load available data from the package
• How to cite this package
• Contributions
• Authors

News of this version (1.2)

• Fixed important issue from IBCF()$predictions_Summary, now the correlation only uses the predicted testing data.
• Now IBCF() function show $yHat that is an average of the prediction values of every partition.
• Now IBCF() function shows in $predicted_Partition all the partitions values predicted (before was $Predictions and was changed to not confuse with $yHat).
• Now IBCF() function shows in $observed all the response values from the DataSet.
• Now IBCF() function shows in $Data.Obs_Pred all the response and predicted values from the DataSet.
• Now IBCF.Years() function shows in $predicted all the values predicted.
• Now IBCF.Years() function shows in $observed the response values for all the years from the Traits selected in Traits.testing.
• Now CV.RandomPart() shows the lenght in every partition in $CrossValidation_length.
• Now CV.RandomPart() admits Traits.testing to only use a PTesting percentage defined of the traits specified in the parameter to be used to fit the model.

See the last updates in NEWS.

Instructions for proper implementation

Installation

To complete installation of dev version of IBCF.MTME from GitHub, you must have previously installed the devtools package.

install.packages('devtools')
devtools::install_github('frahik/IBCF.MTME')

Load the package

library(IBCF.MTME)

Example of Cross-validation with IBCF.MTME

Load available data from other package

library(BGLR)
data(wheat)

Generate a new data set in tidy data form

pheno <- data.frame(ID = gl(n = 599, k = 1, length = 599*4),
                    Response = as.vector(wheat.Y),
                    Env = paste0('Env', gl(n = 4, k = 599)))

head(pheno)

##   ID   Response  Env
## 1  1  1.6716295 Env1
## 2  2 -0.2527028 Env1
## 3  3  0.3418151 Env1
## 4  4  0.7854395 Env1
## 5  5  0.9983176 Env1
## 6  6  2.3360969 Env1

Generate 10 partitions to do cross-validation

CrossV <- CV.RandomPart(pheno, NPartitions = 10, PTesting = 0.25, Set_seed = 123)

Fitting the predictive model

pm <- IBCF(CrossV)

Show some results

summary(pm)

##   Trait_Env Pearson SE_Cor   MSEP SE_MSEP
## 1     _Env1 -0.1307 0.0230 1.9010  0.0452
## 2     _Env2  0.6859 0.0092 0.5454  0.0118
## 3     _Env3  0.6116 0.0164 0.6284  0.0234
## 4     _Env4  0.3068 0.0257 1.0640  0.0495

par(mai = c(2, 1, 1, 1))
plot(pm, select = 'Pearson')

plot(pm, select = 'MSEP')

Example of Years prediction with IBCF.Years Function

Loading your data

load('DataExample.RData')
head(Data.Example)

##   Years Gids  Trait Response
## 1  2014    1 Trait1 15.14401
## 2  2014    2 Trait1 15.67879
## 3  2014    3 Trait1 14.85489
## 4  2014    4 Trait1 13.57002
## 5  2014    5 Trait1 15.01838
## 6  2014    6 Trait1 13.19616

Transforming the data from Tidy data to matrix form

Data.Example <- getMatrixForm(Data.Example, onlyTrait = TRUE)
head(Data.Example)

##   Years Gids   Trait1  Trait10  Trait11  Trait12   Trait2   Trait3
## 1  2014    1 15.14401 18.51428 17.08970 19.16776 16.21435 17.53858
## 2  2014    2 15.67879 18.21569 17.89645 19.94429 15.80614 17.89946
## 3  2014    3 14.85489 17.72576 15.78198 17.53058 14.06164 16.11997
## 4  2014    4 13.57002 18.57009 15.73343 17.49995 14.58312 15.22495
## 5  2014    5 15.01838 18.57348 16.97414 19.03081 14.98192 15.65125
## 6  2014    6 13.19616 16.83588 15.12312 17.39867 15.81264 14.80517
##     Trait4   Trait5   Trait6   Trait7   Trait8   Trait9
## 1 15.51840 17.59132 17.14852 17.04474 17.48970 18.36118
## 2 15.13337 18.36446 17.32734 17.46764 18.08501 18.67266
## 3 15.04329 17.28942 16.50978 16.26685 17.02774 17.05612
## 4 14.93028 16.33687 15.11493 15.06632 17.56798 16.48810
## 5 16.70963 16.81113 17.24170 15.53379 16.07600 16.54047
## 6 14.82150 16.49238 15.37325 14.07796 15.98419 15.84705

Adjust the model

pm <- IBCF.Years(Data.Example, colYears = 1, Years.testing = c('2014', '2015', '2016'), Traits.testing = c('Trait1', 'Trait2', 'Trait3', 'Trait4', "Trait5"))

Show some results

summary(pm)

##              Year_Trait Pearson   MSEP
## 2014_Trait1 2014_Trait1  0.7549 0.4836
## 2014_Trait2 2014_Trait2  0.1562 0.7769
## 2014_Trait3 2014_Trait3  0.6130 0.4164
## 2014_Trait4 2014_Trait4  0.5208 0.6821
## 2014_Trait5 2014_Trait5  0.7587 0.2408
## 2015_Trait1 2015_Trait1  0.8432 0.2987
## 2015_Trait2 2015_Trait2  0.6792 0.5828
## 2015_Trait3 2015_Trait3  0.7944 0.4416
## 2015_Trait4 2015_Trait4  0.7394 0.5425
## 2015_Trait5 2015_Trait5  0.7650 0.4739
## 2016_Trait1 2016_Trait1  0.7690 0.3517
## 2016_Trait2 2016_Trait2  0.7753 0.3818
## 2016_Trait3 2016_Trait3  0.6763 0.5527
## 2016_Trait4 2016_Trait4  0.8157 0.4076
## 2016_Trait5 2016_Trait5  0.8533 0.2779

par(mai = c(3, 1, 1, 1))
barplot(pm, las = 2)

barplot(pm, select = 'MSEP', las = 2)

Load available data from the package

You can use the data sets in the package to test the functions

library(IBCF.MTME)
data('Wheat_IBCF')

head(Wheat_IBCF)

##       GID Trait    Env   Response
## 1 6569128    DH Bed2IR -17.565895
## 2 6688880    DH Bed2IR  -4.565895
## 3 6688916    DH Bed2IR  -3.565895
## 4 6688933    DH Bed2IR  -4.565895
## 5 6688934    DH Bed2IR  -7.565895
## 6 6688949    DH Bed2IR  -7.565895

data('Year_IBCF')

head(Year_IBCF)

##   Years Gids Trait Response
## 1  2014    1    T1 5.144009
## 2  2014    2    T1 5.678792
## 3  2014    3    T1 4.854895
## 4  2014    4    T1 3.570019
## 5  2014    5    T1 5.018380
## 6  2014    6    T1 3.196160

Citation

First option, by the article paper

(Comming soon)

Second option, by the manual package

citation('IBCF.MTME')

## 
## To cite package 'IBCF.MTME' in publications use:
## 
##   Francisco Javier Luna-Vazquez, Osval Antonio Montesinos-Lopez,
##   Abelardo Montesinos-Lopez and Jose Crossa (2018). IBCF.MTME:
##   Item Based Collaborative Filtering for Multi-Trait and
##   Multi-Environment Data. R package version 1.2-5.
##   https://CRAN.R-project.org/package=IBCF.MTME
## 
## A BibTeX entry for LaTeX users is
## 
##   @Manual{,
##     title = {IBCF.MTME: Item Based Collaborative Filtering for Multi-Trait and
## Multi-Environment Data},
##     author = {Francisco Javier Luna-Vazquez and Osval Antonio Montesinos-Lopez and Abelardo Montesinos-Lopez and Jose Crossa},
##     year = {2018},
##     note = {R package version 1.2-5},
##     url = {https://CRAN.R-project.org/package=IBCF.MTME},
##   }

Contributions

If you have any suggestions or feedback, I would love to hear about it. Feel free to report new issues in this link, also if you want to request a feature/report a bug, or make a pull request if you can contribute.

Authors

• Francisco Javier Luna-Vázquez (Author, Maintainer)
• Osval Antonio Montesinos-López (Author)
• Abelardo Montesinos-López (Author)
• José Crossa (Author)