Common Task Framework

About

A system to compare algorithms made by several users on a common set of datasets.

Leaderboard

Click here to see the leaderboard

Environment

2.1GHz 6-core CPU
24GB RAM
40GB Hard disk space available

R version 3.3.1 (2016-06-21)
Platform: x86_64-redhat-linux-gnu (64-bit)
Running under: CentOS Linux 7 (Core)
9383 R packages installed

Internet access to CRAN, Bioconductor, GitHub only for installing additional packages.

Datasets

Data structure example:

data_df
       lon       lat      co2
 -174.9372 -38.10415 395.0963
 -179.0481 -22.81784 392.5952
  172.3957  16.30791 395.6466
  172.4112  16.29490 393.4368
  172.3898  16.35039 393.9456

pred_df
       lon        lat
 -94.65960  -8.412008
 -13.03596 -39.143768
 144.14600 -12.843467
  59.03585  46.178749
 128.56120   1.963521

Example function

Input:

spat_interp <- function(data_df,pred_df) {
    library(LatticeKrig)
    obj <- LatticeKrig(data_df[,1:2], data_df[,3],nlevel=3,NC=8)
    out.pred <- predict.LKrig(obj, xnew= pred_df[,1:2])
    J <- LKrig.sim.conditional(obj, x.grid = as.matrix(pred_df[,1:2]),M=30)
    LTK_pred <- out.pred
    LTK_se <- sqrt(apply(J[[3]],1,sd)^2 + obj$sigma.MLE)
    list(mu=LTK_pred,se=LTK_se)
}

Output:

results$mu
         [,1]
[1,] 402.3051
[2,] 398.2299
[3,] 397.4613
[4,] 397.6921
[5,] 396.5963

results$se
[1] 1.813732 2.261943 1.808671 1.812388 1.804914

Results summary:

results_summary
  user       date method  dataset comptime      rmspe          pe  cov90
1  bob 2016-11-15    LTK oco2lite       NA 0.07042912 -0.07042912 0.7178

Submit a Task

Your name:
Email address:
Function name:
Dataset:
Difficulty:

Upload a '.R' file containing a single function that accepts as input the two data frames data_df and pred_df, and returns a list of mu and se.