generate pdf version of tutorial

.travis.yml

@@ -22,6 +22,7 @@ addons:
       - g++-4.9
       - gfortran-4.9
       - libgmp-dev
+      - pandoc
 
 before_install:
   - mkdir $HOME/bin
@@ -38,7 +39,7 @@ install:
   - Rscript r_packs_install.R
 
 script:
-  - ./build || ./build || ./build || ./build
+  - ./build
 
 after_success:
     - git checkout gh-pages
@@ -48,6 +49,7 @@ after_success:
     - echo "https://$GH_TOKEN:@github.com" >> .git/credentials
     - git config push.default matching
     - git add devel
+    - git add mlr-tutorial.pdf
     - git commit -a -m "update auto-generated tutorial pages [ci skip]"
     - git push
 

README.md

@@ -29,8 +29,8 @@ Install dependencies:
 
 ### Include images
 If you want to include an additional image in file `pic.png`:  
-* Put this file in subfolder `images/`.
-* Add a symlink in directory `custom_theme/img/`: `pic.png -> ../../images/pic.png`.
+* Put this file in subfolder `img/`.
+* Add a symlink in directory `custom_theme/img/`: `pic.png -> ../../img/pic.png`.
 * When including the image in the R markdown link to `img/pic.png`:  
   `![alt text](img/pic.png "Image Title")`
 

build

@@ -77,14 +77,14 @@ setSeed = function(x, algo = "crc32") {
 }
 
 # worker function
-knitIt = function(f) {
+knitIt = function(f, dev) {
   messagef("Knitting file '%s' ...", f)
   lines = readLines(file.path(src, f))
   for (macro in macros)
     lines = str_replace_all(lines, macro$pattern, macro$replacement)
 
   cache.path = paste0(file.path("cache", f), .Platform$file.sep)
-  opts_chunk$set(cache = TRUE, cache.path = cache.path, dev = "svg", error = FALSE, comment = "#>", collapse = TRUE)
+  opts_chunk$set(cache = TRUE, cache.path = cache.path, dev = dev, error = FALSE, comment = "#>", collapse = TRUE)
   setSeed(f)
   knit(
     text = lines,
@@ -116,8 +116,6 @@ suppressPackageStartupMessages({
 
 # turn warnings to errors, we don't want to miss them
 options(warn = 2L)
-# embed images in HTML
-opts_knit$set(upload.fun = image_uri)
 # number of spaces to indent the code is 2 (for chunks with tidy=TRUE)
 opts_chunk$set(tidy.opts=list(indent=2, width.cutoff=80))
 
@@ -126,18 +124,20 @@ if (!isDirectory("docs"))
   dir.create("docs")
 
 # do lapply for now. with mclapply it seems unclear how to stop best on error. this was 100% broken before
-dummy = lapply(list.files(src, pattern = "\\.Rmd$"), knitIt)
-
-# build docs with mkdocs
-ok = system3("mkdocs", "build")
+dummy = lapply(list.files(src, pattern = "\\.Rmd$"), knitIt, "pdf")
+message("Building PDF version...")
+ok = system3("./build-pdf.py")
 if (ok$exit.code == 0L) {
-  message("Build successfull!")
+  # embed images in HTML
+  opts_knit$set(upload.fun = image_uri)
+  message("Building HTML version...")
+  dummy = lapply(list.files(src, pattern = "\\.Rmd$"), knitIt, "svg")
+  ok = system3("mkdocs", "build")
+  message("Build successful!")
   message("Now zip everything!")
   setwd("devel/html/")
   suppressAll(zip("../mlr_tutorial.zip", files = "."))
-  setwd("../")
-  # clean up
-  unlink("figure", recursive = TRUE)
+  setwd("../../")
   exit(0L)
 }
 

build-pdf.py

@@ -0,0 +1,25 @@
+#!/usr/bin/env python
+import os
+import re
+import sys
+
+os.chdir("docs")
+os.system("echo -e '---\ntitle: mlr Tutorial\n---' > ../mlr-tutorial.md")
+os.system("for f in `grep -o \"[^']\+\.md\" ../mkdocs.yml`; do (cat \"${f}\"; echo) >> ../mlr-tutorial.md; done")
+os.chdir("..")
+
+def link_fixer(match):
+  file = match.group(1)
+  with open(os.path.join("docs", file), 'r') as fd:
+    return "(" + re.sub(' ', '-', re.sub(' ', '', fd.readline().rstrip(), 1)).lower() + ")"
+
+with open('mlr-tutorial.md', 'r+') as file:
+  data = file.read()
+  out = re.sub(r'\(([a-zA-Z0-9_]+\.md)\)', link_fixer, data)
+  file.seek(0)
+  file.write(out)
+  file.truncate()
+
+retval = os.system("pandoc --latex-engine=xelatex --variable colorlinks=\"true\" --listings -H latex-setup.tex --toc -f markdown+grid_tables+table_captions-implicit_figures -o mlr-tutorial.pdf mlr-tutorial.md")
+if retval != 0:
+  sys.exit(1)

custom_theme/img/mlrLogo_white_88x40.png

@@ -1 +1 @@
-../../images/mlrLogo_white_88x40.png
+../../img/mlrLogo_white_88x40.png

custom_theme/img/nested_resampling.png

@@ -1 +1 @@
-../../images/nested_resampling.png
+../../img/nested_resampling.png

custom_theme/img/resampling.png

@@ -1 +1 @@
-../../images/resampling.png
+../../img/resampling.png

devel/html/index.html

@@ -437,5 +437,5 @@ <h4 class="modal-title" id="exampleModalLabel">Search</h4>
 
 <!--
 MkDocs version : 0.15.3
-Build Date UTC : 2016-08-25 19:03:40.119325
+Build Date UTC : 2016-08-29 05:12:46.110878
 -->

devel/html/learner/index.html

@@ -529,9 +529,6 @@ <h2 id="modifying-a-learner">Modifying a learner</h2>
 Here are some examples.</p>
 <pre><code class="r">## Change the ID
 surv.lrn = setId(surv.lrn, &quot;CoxModel&quot;)
-#&gt; Warning: 'setId' is deprecated.
-#&gt; Use 'setLearnerId' instead.
-#&gt; See help(&quot;Deprecated&quot;)
 surv.lrn
 #&gt; Learner CoxModel from package survival
 #&gt; Type: surv

devel/html/measures/index.html

@@ -933,20 +933,6 @@ <h3 id="regression">Regression</h3>
 <td align="left"></td>
 </tr>
 <tr>
-<td align="left"><strong><a href="http://www.rdocumentation.org/packages/mlr/functions/measures.html">mape</a></strong> <br />Mean absolute percentage error</td>
-<td align="center">X</td>
-<td align="right">0</td>
-<td align="right">Inf</td>
-<td align="center">X</td>
-<td align="center">X</td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="left"><a href="http://www.rdocumentation.org/packages/mlr/functions/aggregations.html">test.mean</a></td>
-<td align="left">Defined as the abs(truth_i - response_i)/truth_i. Won't work if any truth value is equal to zero. In this case the output will be NA.</td>
-</tr>
-<tr>
 <td align="left"><strong><a href="http://www.rdocumentation.org/packages/mlr/functions/measures.html">medae</a></strong> <br />Median of absolute errors</td>
 <td align="center">X</td>
 <td align="right">0</td>
@@ -989,20 +975,6 @@ <h3 id="regression">Regression</h3>
 <td align="left"></td>
 </tr>
 <tr>
-<td align="left"><strong><a href="http://www.rdocumentation.org/packages/mlr/functions/measures.html">rae</a></strong> <br />Relative absolute error</td>
-<td align="center">X</td>
-<td align="right">0</td>
-<td align="right">Inf</td>
-<td align="center">X</td>
-<td align="center">X</td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="left"><a href="http://www.rdocumentation.org/packages/mlr/functions/aggregations.html">test.mean</a></td>
-<td align="left">Defined as sum_of_absolute_errors / mean_absolute_deviation. Undefined for single instances and when every truth value is identical. In this case the output will be NA.</td>
-</tr>
-<tr>
 <td align="left"><strong><a href="http://www.rdocumentation.org/packages/mlr/functions/measures.html">rmse</a></strong> <br />Root mean square error</td>
 <td align="center">X</td>
 <td align="right">0</td>
@@ -1017,20 +989,6 @@ <h3 id="regression">Regression</h3>
 <td align="left">The RMSE is aggregated as sqrt(mean(rmse.vals.on.test.sets^2)). If you don't want that, you could also use <code>test.mean</code>.</td>
 </tr>
 <tr>
-<td align="left"><strong><a href="http://www.rdocumentation.org/packages/mlr/functions/measures.html">rrse</a></strong> <br />Root relative squared error</td>
-<td align="center">X</td>
-<td align="right">0</td>
-<td align="right">Inf</td>
-<td align="center">X</td>
-<td align="center">X</td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="center"></td>
-<td align="left"><a href="http://www.rdocumentation.org/packages/mlr/functions/aggregations.html">test.mean</a></td>
-<td align="left">Defined as sqrt (sum_of_squared_errors / total_sum_of_squares). Undefined for single instances and when every truth value is identical. In this case the output will be NA.</td>
-</tr>
-<tr>
 <td align="left"><strong><a href="http://www.rdocumentation.org/packages/mlr/functions/measures.html">rsq</a></strong> <br />Coefficient of determination</td>
 <td align="center"></td>
 <td align="right">1</td>

devel/html/multilabel/index.html

@@ -550,13 +550,13 @@ <h2 id="performance">Performance</h2>
 #&gt; multilabel.subset01  multilabel.hamloss      multilabel.acc 
 #&gt;           0.8663633           0.2049471           0.4637509 
 #&gt;       multilabel.f1         timepredict 
-#&gt;           0.5729926           1.4540000
+#&gt;           0.5729926           1.0810000
 
 listMeasures(&quot;multilabel&quot;)
-#&gt;  [1] &quot;multilabel.f1&quot;       &quot;multilabel.subset01&quot; &quot;multilabel.tpr&quot;     
-#&gt;  [4] &quot;multilabel.ppv&quot;      &quot;multilabel.acc&quot;      &quot;timeboth&quot;           
-#&gt;  [7] &quot;timepredict&quot;         &quot;multilabel.hamloss&quot;  &quot;featperc&quot;           
-#&gt; [10] &quot;timetrain&quot;
+#&gt;  [1] &quot;multilabel.ppv&quot;      &quot;timepredict&quot;         &quot;multilabel.hamloss&quot; 
+#&gt;  [4] &quot;multilabel.f1&quot;       &quot;featperc&quot;            &quot;multilabel.subset01&quot;
+#&gt;  [7] &quot;timeboth&quot;            &quot;timetrain&quot;           &quot;multilabel.tpr&quot;     
+#&gt; [10] &quot;multilabel.acc&quot;
 </code></pre>
 
 <h2 id="resampling">Resampling</h2>
@@ -573,7 +573,7 @@ <h2 id="resampling">Resampling</h2>
 #&gt; multilabel.hamloss.aggr: 0.22
 #&gt; multilabel.hamloss.mean: 0.22
 #&gt; multilabel.hamloss.sd: 0.01
-#&gt; Runtime: 5.76838
+#&gt; Runtime: 4.37393
 
 r = resample(learner = lrn.rFerns, task = yeast.task, resampling = rdesc, show.info = FALSE)
 r
@@ -583,7 +583,7 @@ <h2 id="resampling">Resampling</h2>
 #&gt; multilabel.hamloss.aggr: 0.47
 #&gt; multilabel.hamloss.mean: 0.47
 #&gt; multilabel.hamloss.sd: 0.00
-#&gt; Runtime: 0.540231
+#&gt; Runtime: 0.404812
 </code></pre>
 
 <h2 id="binary-performance">Binary performance</h2>

devel/html/preproc/index.html

@@ -767,7 +767,7 @@ <h3 id="creating-the-preprocessing-wrapper">Creating the preprocessing wrapper</
 #&gt; mse.aggr: 20.62
 #&gt; mse.mean: 20.62
 #&gt; mse.sd: 8.53
-#&gt; Runtime: 0.132104
+#&gt; Runtime: 0.103239
 
 lrn = makeLearner(&quot;regr.nnet&quot;, trace = FALSE, decay = 1e-02)
 r = resample(lrn, bh.task, resampling = rdesc, show.info = FALSE)
@@ -778,7 +778,7 @@ <h3 id="creating-the-preprocessing-wrapper">Creating the preprocessing wrapper</
 #&gt; mse.aggr: 55.06
 #&gt; mse.mean: 55.06
 #&gt; mse.sd: 19.40
-#&gt; Runtime: 0.10274
+#&gt; Runtime: 0.0779717
 </code></pre>
 
 <h3 id="joint-tuning-of-preprocessing-and-learner-parameters">Joint tuning of preprocessing and learner parameters</h3>
@@ -848,18 +848,18 @@ <h3 id="joint-tuning-of-preprocessing-and-learner-parameters">Joint tuning of pr
 
 as.data.frame(res$opt.path)
 #&gt;    decay center scale mse.test.mean dob eol error.message exec.time
-#&gt; 1      0   TRUE  TRUE      49.38128   1  NA          &lt;NA&gt;     0.042
-#&gt; 2   0.05   TRUE  TRUE      20.64761   2  NA          &lt;NA&gt;     0.049
-#&gt; 3    0.1   TRUE  TRUE      22.42986   3  NA          &lt;NA&gt;     0.049
-#&gt; 4      0  FALSE  TRUE      96.25474   4  NA          &lt;NA&gt;     0.027
-#&gt; 5   0.05  FALSE  TRUE      14.84306   5  NA          &lt;NA&gt;     0.052
-#&gt; 6    0.1  FALSE  TRUE      16.65383   6  NA          &lt;NA&gt;     0.047
-#&gt; 7      0   TRUE FALSE      40.51518   7  NA          &lt;NA&gt;     0.050
-#&gt; 8   0.05   TRUE FALSE      68.00069   8  NA          &lt;NA&gt;     0.047
-#&gt; 9    0.1   TRUE FALSE      55.42210   9  NA          &lt;NA&gt;     0.050
-#&gt; 10     0  FALSE FALSE      96.25474  10  NA          &lt;NA&gt;     0.027
-#&gt; 11  0.05  FALSE FALSE      56.25758  11  NA          &lt;NA&gt;     0.049
-#&gt; 12   0.1  FALSE FALSE      42.85529  12  NA          &lt;NA&gt;     0.049
+#&gt; 1      0   TRUE  TRUE      49.38128   1  NA          &lt;NA&gt;     0.038
+#&gt; 2   0.05   TRUE  TRUE      20.64761   2  NA          &lt;NA&gt;     0.045
+#&gt; 3    0.1   TRUE  TRUE      22.42986   3  NA          &lt;NA&gt;     0.045
+#&gt; 4      0  FALSE  TRUE      96.25474   4  NA          &lt;NA&gt;     0.023
+#&gt; 5   0.05  FALSE  TRUE      14.84306   5  NA          &lt;NA&gt;     0.049
+#&gt; 6    0.1  FALSE  TRUE      16.65383   6  NA          &lt;NA&gt;     0.043
+#&gt; 7      0   TRUE FALSE      40.51518   7  NA          &lt;NA&gt;     0.044
+#&gt; 8   0.05   TRUE FALSE      68.00069   8  NA          &lt;NA&gt;     0.042
+#&gt; 9    0.1   TRUE FALSE      55.42210   9  NA          &lt;NA&gt;     0.045
+#&gt; 10     0  FALSE FALSE      96.25474  10  NA          &lt;NA&gt;     0.022
+#&gt; 11  0.05  FALSE FALSE      56.25758  11  NA          &lt;NA&gt;     0.044
+#&gt; 12   0.1  FALSE FALSE      42.85529  12  NA          &lt;NA&gt;     0.042
 </code></pre>
 
 <h3 id="preprocessing-wrapper-functions">Preprocessing wrapper functions</h3>

devel/html/resample/index.html

@@ -411,7 +411,7 @@ <h1 id="resampling">Resampling</h1>
 #&gt; cindex.aggr: 0.63
 #&gt; cindex.mean: 0.63
 #&gt; cindex.sd: 0.05
-#&gt; Runtime: 0.0429173
+#&gt; Runtime: 0.0281701
 ## peak a little bit into r
 names(r)
 #&gt;  [1] &quot;learner.id&quot;     &quot;task.id&quot;        &quot;measures.train&quot; &quot;measures.test&quot; 
@@ -485,10 +485,10 @@ <h1 id="resampling">Resampling</h1>
 #&gt; [Resample] subsampling iter: 3
 #&gt; [Resample] subsampling iter: 4
 #&gt; [Resample] subsampling iter: 5
-#&gt; [Resample] Result: dunn.test.mean=0.274,db.test.mean=0.51,timetrain.test.mean=0.0008
+#&gt; [Resample] Result: dunn.test.mean=0.274,db.test.mean=0.51,timetrain.test.mean=0.001
 r$aggr
 #&gt;      dunn.test.mean        db.test.mean timetrain.test.mean 
-#&gt;           0.2738893           0.5103655           0.0008000
+#&gt;           0.2738893           0.5103655           0.0010000
 </code></pre>
 
 <h2 id="stratified-resampling">Stratified resampling</h2>