Running parallel jobs on Aster with R and toaster

data-set-model-and-K-folds.R

    library(toaster)  
      
    close(conn)  
    conn = odbcDriverConnect(connection="driver={Aster ODBC Driver};server=10.xx.xx.xx;port=2406;database=dallas;uid=beehive;pwd=beehive",  
                             interpretDot=TRUE)  
      
    dallasPermitsTableInfo = getTableSummary(conn, "dallasbuildingpermits")  
      
    getNumericColumns(dallasPermitsTableInfo)  

execute-cv-par-run.R

    fit.folds = foreach(fold = 1:12, .packages=c('RODBC','toaster')) %dopar% {  
      parConn = odbcReConnect(conn)  
      fitted.model = computeLm(parConn, "dallasbuildingpermits", value ~ area + lon + lat, tableInfo = dallasPermitsTableInfo,  
                        where = paste("date_part('month',issued)::int != ", fold))  
      close(parConn)  
      fitted.model  
    }

execute-cv-par-setup.R

library(doParallel)  
  
cl = makeCluster(6)  
registerDoParallel(cl, cores=6)

execute-cv-seq-par.R

    library(foreach)  
      
    # running cross-validation sequantially with foreach function  
    fit.folds = foreach(fold = 1:12, .packages=c('RODBC', 'toaster')) %do% {  
      fitted.model = computeLm(conn, "dallasbuildingpermits", value ~ area + lon + lat, tableInfo = dallasPermitsTableInfo,  
                        where = paste("date_part('month',issued)::int != ", fold))  
    }  

execute-cv-seq.R

fit.folds = list()  
for(fold in 1:12) {  
  fit.folds[[fold]] = computeLm(conn, "dallasbuildingpermits", value ~ area + lon + lat, tableInfo = dallasPermitsTableInfo,  
                    where = paste("date_part('month',issued)::int !=", fold))  
} 

install-packages.R

install.packages("toaster", "parallel", "iterators", "doParallel", "foreach")

k-folds-create.sql

    SELECT DATE_PART('month',issued)::int fold, COUNT(*) fold_size   
      FROM public.dallasbuildingpermits   
     GROUP BY 1   
     ORDER BY 1  

k-folds-select.sql

SELECT COUNT(*)  
  FROM public.dallasbuildingpermits  
WHERE DATE_PART('month',issued)::int != 6 

linreg-formula-in.R

value ~ area + lon + lat

linreg-with-toaster.R

fit.all = computeLm(conn, "dallasbuildingpermits", value ~ area + lon + lat, tableInfo = dallasPermitsTableInfo)