Update WrapRec for 2020

Appears that there are some issues with us supporting WrapRec that need to be resolved before it can be used for teaching:

1. Update documentation on Brightspace

   from datasets import CS4065_Dataset
   from wraprec import PyWrapRec
   
   config_file_path = CS4065_Dataset.get_wraprec_sample_data()

   should be replaced with

   from datasets import CS4065_Dataset
   from wraprec import PyWrapRec
   
   cs4065_dataset = CS4065_Dataset()
   config_file_path = cs4065_dataset.get_wraprec_sample_data()

2. Executing !mono /home/student/wraprec/WrapRec.exe /home/student/data/cs4065/wraprec_sample_data3/sample.xml inside the notebook using the default sample.xml file produces the following output:

   Case 1 of 4:
   ----------------------------------------
   Loading DataContainer 'ml100k'...
   Loading data 'ml100k-all' into container...
   Setuping evaluation context 'eval'...
   Setuping model 'mf'...
   Error in expriment 'ml100k', model 'mf', split 'ml100k-cv-fold1':
   Cannot resolve MmlRecommender: Could not load file or assembly 'MyMediaLite.dll' or one of its dependencies
     at (wrapper managed-to-native) System.Reflection.Assembly:LoadFrom (string,bool)
     at System.Reflection.Assembly.LoadFrom (System.String assemblyFile) <0x7f51a9926860 + 0x0000e> in <filename unknown>:0 
     at WrapRec.Utils.Helpers.ResolveType (System.String typeName) <0x4085a920 + 0x00074> in <filename unknown>:0 
     at WrapRec.Models.MmlRecommender.Setup () <0x4086be30 + 0x0005f> in <filename unknown>:0 
     at WrapRec.Models.MmlRecommender.Setup () <0x4086be30 + 0x00583> in <filename unknown>:0 
     at WrapRec.Core.Experiment.Setup () <0x408644f0 + 0x0010b> in <filename unknown>:0 
     at WrapRec.Core.ExperimentManager.RunSingleExperiment (WrapRec.Core.Experiment e) <0x40863ef0 + 0x000a7> in <filename unknown>:0 
   
   Case 2 of 4:
   ----------------------------------------
   <OTHER 3 CASES>
   Experiments are executed: 0 succeeded, 4 failed.
   Results are stored in /home/student/data/cs4065/wraprec_sample_data3/results
   Total running time: 0 seconds.

   When using the python wrapper the students are provided with (PyWrapRec) the error messages are not shown to the user and the program executes producing the output csv files. It is unclear whether the above error messages can be safely ignored or should be fixed.

3. When replacing the second model in sample.xml with

   <model id="bmf" class="WrapRec.Models.MmlRecommender">
     <parameters ml-class="MyMediaLite.dll:MyMediaLite.RatingPrediction.ItemKNN" K="3" />
   </model>

   executing the command via mono (as above) completes (with 0 successes as above) and outputs are produced. Executing the command via the python wrapper leads to the notebook hanging.

4. When replacing the second model in sample.xml with

   <model id="libfm" class="WrapRec.Models.LibFmWrapper"><parameters libFmPath="libfm.net.exe" task="r" dim="1-1-10" method="sgd" iter="30" learn_rate="0.02"/></model>

   and

   <experiment id="ml100k" models="mf,bmf" splits="ml100k-cv" evalContext="eval" />

   with

   <experiment id="ml100k" models="mf,libfm" splits="ml100k-cv" evalContext="eval" />

   the following output is produced:

   Case 2 of 4:
   ----------------------------------------
   Setuping evaluation context 'eval'...
   Setuping model 'libfm'...
   Calculating split 'ml100k-cv-fold1' statistics...
   Calculating dataContainer 'ml100k' statistics...
   
   Experiment Id: ml100k
   Split Id: ml100k-cv-fold1
   Model Id: libfm
   Model Parameteres:
   task:r dim:1,1,10 method:sgd iter:30 learn_rate:0.02
   
   Training...
   Creating LibFm train and test files...
   Training and testing with LibFm...
   Running process:
    libfm.net.exe -task r -dim 1,1,10 -method sgd -iter 30 -learn_rate 0.02 -save_model train.model -train train.libfm -test test.libfm -out test.out
   Error in expriment 'ml100k', model 'libfm', split 'ml100k-cv-fold1':
   ApplicationName='libfm.net.exe', CommandLine='-task r -dim 1,1,10 -method sgd -iter 30 -learn_rate 0.02 -save_model train.model -train train.libfm -test test.libfm -out test.out', CurrentDirectory='', Native error= Cannot find the specified file
     at System.Diagnostics.Process.Start_noshell (System.Diagnostics.ProcessStartInfo startInfo, System.Diagnostics.Process process) <0x41be2f10 + 0x00fe3> in <filename unknown>:0 
     at System.Diagnostics.Process.Start_common (System.Diagnostics.ProcessStartInfo startInfo, System.Diagnostics.Process process) <0x41be2d60 + 0x000db> in <filename unknown>:0 
     at System.Diagnostics.Process.Start () <0x41be2ce0 + 0x0004b> in <filename unknown>:0 
     at (wrapper remoting-invoke-with-check) System.Diagnostics.Process:Start ()
     at WrapRec.Models.LibFmWrapper+<>c__DisplayClass11.<Train>b__e () <0x41be2bf0 + 0x0001b> in <filename unknown>:0 
     at MyMediaLite.Wrap.MeasureTime (System.Action t) <0x41bca510 + 0x00023> in <filename unknown>:0 
     at WrapRec.Models.LibFmWrapper.Train (WrapRec.Data.Split split) <0x41bdee10 + 0x0082f> in <filename unknown>:0 
     at WrapRec.Core.Experiment.<Run>b__0 () <0x41bdeda0 + 0x0001e> in <filename unknown>:0 
     at MyMediaLite.Wrap.MeasureTime (System.Action t) <0x41bca510 + 0x00023> in <filename unknown>:0 
     at WrapRec.Core.Experiment.Run () <0x41bdebb0 + 0x000c3> in <filename unknown>:0 
     at WrapRec.Core.ExperimentManager.RunSingleExperiment (WrapRec.Core.Experiment e) <0x41bcaf80 + 0x0018c> in <filename unknown>:0 

   and the training does not complete.

5. For the cases where the training does complete the results pandas DataFrame contains partially wrong values (row 3):