fixed regression/classification round;line up repeated cv;removed total in cv matrix;

3 files changed, 118 insertions, 112 deletions

diff --git a/views/neighbors.haml b/views/neighbors.haml
index 69034b3..70bf4b8 100644
--- a/views/neighbors.haml
+++ b/views/neighbors.haml
@@ -10,6 +10,7 @@
     %ul.nav.nav-tabs.nav-justified{:id=>"neighborTabs", :role=>"tablist"}
       / each model a tab head ;
       - @models.each_with_index do |model,i|
+        - @type = model.model.class.to_s.match("Classification") ? "Classification" : "Regression"
         %li{:class => ("active" if i == 0)}
           %a{:href => "#results_#{i+1}", :id => "linkTab#{i+1}", data: {toggle:"tab"}}
             = "#{model.endpoint} (#{model.species})"
@@ -31,7 +32,8 @@
                 headerTemplate: '{content} {icon}',
                 widgets: ['zebra', 'columns', 'uitheme'],
                 headers: {0: {sorter: false}, 3: {sorter: false}},
-                sortList: [[2,1]]
+                sortList: [[2,1]],
+                widthFixed: false
               });
             });
           - if prediction[:neighbors].size > 0
@@ -59,16 +61,20 @@
                 %tbody
                 - prediction[:neighbors].each_with_index do |neighbor,count|
                   %tr
-                    %td{:style =>"vertical-align:middle;padding-left:1em;"}
+                    / Compound
+                    %td{:style =>"vertical-align:middle;padding-left:1em;width:50%;"}
                       /%a.btn.btn-link{:href => "#details#{j+1}", data: { toggle: "modal", remote: to("/prediction/#{CGI.escape(neighbor[0].to_s)}/details"), :id=>"link#{j+1}#{count}"}}
                       %p= Compound.find(neighbor[0]).svg
                       %p= Compound.find(neighbor[0]).smiles
-                    %td{:style =>"vertical-align:middle;padding-left:1em;"}
-                      -#= neighbor[2].collect{|n| '%.2e' % n}.join ", "
-                      = neighbor[2].join ", "
-                    %td{:style =>"vertical-align:middle;padding-left:1em;"}
+                    / Measured Activity
+                    %td{:style =>"vertical-align:middle;padding-left:1em;width:20%;"}
+                      //= neighbor[2].collect{|n| '%.2e' % n}.join ", "
+                      //= neighbor[2].join ", "
+                      = (@type == "Regression") ? neighbor[2].collect{|n| n.round(2)}.join(", ") : neighbor[2].join(", ")
+                    / Similarity
+                    %td{:style =>"vertical-align:middle;padding-left:1em;width:20%;"}
                       / TODO differentiate between no neighbors found and compound found in dataset, display neighbors for compounds in dataset?
-                      = neighbor[1] != nil ? neighbor[1] : "Not enough similar compounds in training dataset." 
+                      = neighbor[1] != nil ? neighbor[1].round(2) : "Not enough similar compounds in training dataset." 
 
           - else
             %span.btn.btn-default.disabled Not enough similar compounds in training dataset
diff --git a/views/predict.haml b/views/predict.haml
index 0e6b3cc..1d75723 100644
--- a/views/predict.haml
+++ b/views/predict.haml
@@ -144,7 +144,7 @@
               %br
               = "Algorithm:\tLAZAR"
               %br
-              - model.classification? ? type = "classification" : type = "regression"
+              - model.classification? ? type = "Classification" : type = "Regression"
               = "Type:\t"
               = type
               %br
@@ -155,108 +155,107 @@
               = "Training compounds:\t"
               = training_dataset.compounds.size
 
-
               %p
               %b Validation (repeated):
-              %p
-              - model.crossvalidations.each do |crossvalidation|
-                - cv = OpenTox::CrossValidation.find crossvalidation.id
-                = "Num folds:\t"
-                = cv.folds
-                %br
-                = "Num instances:\t"
-                = cv.nr_instances
-                %br
-                = "Num unpredicted"
-                = cv.nr_unpredicted
-                - if model.classification? 
-                  %br
-                  = "Accuracy:\t"
-                  = cv.accuracy.round(3) if cv.accuracy
-                  %br
-                  = "Weighted Accuracy:\t"
-                  = cv.weighted_accuracy.round(3) if cv.weighted_accuracy
-                  %br
-                  = "True positive rate:\t"
-                  = cv.true_rate["active"].round(3) if cv.true_rate["active"]
-                  %br
-                  = "True negative rate:\t"
-                  = cv.true_rate["inactive"].round(3) if cv.true_rate["inactive"]
-                  %br 
-                  = "Positive predictive value:\t"
-                  = cv.predictivity["active"].round(3) if cv.predictivity["active"]
-                  %br
-                  = "Negative predictive value:\t"
-                  = cv.predictivity["inactive"].round(3) if cv.predictivity["inactive"]
-                  %p
-                  %b Confusion Matrix:
-                  %table.table.table-condensed.table-borderless{:style=>"width:20%;"}
-                    %tbody
-                      %tr
-                        %td
-                        %td
-                        %td 
-                          %b actual
-                        %td
-                        %td
-                      %tr
-                        %td
-                        %td
-                        %td active
-                        %td inactive
-                        %td total
-                      %tr
-                        %td 
-                          %b predicted
-                        %td active
-                        %td 
-                          =cv.confusion_matrix[0][0]
-                        %td 
-                          =cv.confusion_matrix[0][1]
-                        %td 
-                          =cv.confusion_matrix[0][0]+cv.confusion_matrix[0][1]
-                      %tr
-                        %td
-                        %td inactive
-                        %td 
-                          =cv.confusion_matrix[1][0]
-                        %td 
-                          =cv.confusion_matrix[1][1]
-                        %td 
-                          =cv.confusion_matrix[1][0]+cv.confusion_matrix[1][1]
-                      %tr
-                        %td
-                        %td total
-                        %td
-                          =cv.confusion_matrix[0][0]+cv.confusion_matrix[1][0]
-                        %td
-                          =cv.confusion_matrix[0][1]+cv.confusion_matrix[1][1]
-                        %td
-                          -#= "Confusion Matrix:\t"
-                          -#= cv.confusion_matrix
-                - if model.regression?
-                  %br
-                  = "Root mean squared error:\t"
-                  = cv.rmse.round(3) if cv.rmse
-                  %br
-                  = "Weighted root mean squared error:\t"
-                  = cv.weighted_rmse.round(3) if cv.weighted_rmse
-                  %br
-                  = "Mean absolute error:\t"
-                  = cv.mae.round(3) if cv.mae
-                  %br
-                  = "Weighted mean absolute error:\t"
-                  = cv.weighted_mae.round(3) if cv.weighted_mae
-                  %br 
-                  = "R square:\t"
-                  = cv.r_squared.round(3) if cv.r_squared
-                  /%br
-                  /= "Correlation plot"
-                  /= cv.correlation_plot
-                  /%br
-                  /= "Confidence plot:"
-                  /= cv.confidence_plot
-                %hr
+              %div.row{:style=>"background-color:#f5f5f5;"}
+                - model.crossvalidations.each do |crossvalidation|
+                  %span.col-xs-4.col-sm-4.col-md-4.col-lg-4
+                    - cv = OpenTox::CrossValidation.find crossvalidation.id
+                    = "Num folds:\t"
+                    = cv.folds
+                    %br
+                    = "Num instances:\t"
+                    = cv.nr_instances
+                    %br
+                    = "Num unpredicted"
+                    = cv.nr_unpredicted
+                    - if model.classification? 
+                      %br
+                      = "Accuracy:\t"
+                      = cv.accuracy.round(3) if cv.accuracy
+                      %br
+                      = "Weighted Accuracy:\t"
+                      = cv.weighted_accuracy.round(3) if cv.weighted_accuracy
+                      %br
+                      = "True positive rate:\t"
+                      = cv.true_rate["active"].round(3) if cv.true_rate["active"]
+                      %br
+                      = "True negative rate:\t"
+                      = cv.true_rate["inactive"].round(3) if cv.true_rate["inactive"]
+                      %br 
+                      = "Positive predictive value:\t"
+                      = cv.predictivity["active"].round(3) if cv.predictivity["active"]
+                      %br
+                      = "Negative predictive value:\t"
+                      = cv.predictivity["inactive"].round(3) if cv.predictivity["inactive"]
+                      %p
+                      %b Confusion Matrix:
+                      %table.table.table-condensed.table-borderless{:style=>"width:20%;"}
+                        %tbody
+                          %tr
+                            %td
+                            %td
+                            %td 
+                              %b actual
+                            %td
+                            %td
+                          %tr
+                            %td
+                            %td
+                            %td active
+                            %td inactive
+                            %td total
+                          %tr
+                            %td 
+                              %b predicted
+                            %td active
+                            %td 
+                              =cv.confusion_matrix[0][0]
+                            %td 
+                              =cv.confusion_matrix[0][1]
+                            %td 
+                              =cv.confusion_matrix[0][0]+cv.confusion_matrix[0][1]
+                          %tr
+                            %td
+                            %td inactive
+                            %td 
+                              =cv.confusion_matrix[1][0]
+                            %td 
+                              =cv.confusion_matrix[1][1]
+                            %td 
+                              =cv.confusion_matrix[1][0]+cv.confusion_matrix[1][1]
+                          -#%tr
+                            %td
+                            %td total
+                            %td
+                              =cv.confusion_matrix[0][0]+cv.confusion_matrix[1][0]
+                            %td
+                              =cv.confusion_matrix[0][1]+cv.confusion_matrix[1][1]
+                            %td
+                              -#= "Confusion Matrix:\t"
+                              -#= cv.confusion_matrix
+                    - if model.regression?
+                      %br
+                      = "Root mean squared error:\t"
+                      = cv.rmse.round(3) if cv.rmse
+                      %br
+                      = "Weighted root mean squared error:\t"
+                      = cv.weighted_rmse.round(3) if cv.weighted_rmse
+                      %br
+                      = "Mean absolute error:\t"
+                      = cv.mae.round(3) if cv.mae
+                      %br
+                      = "Weighted mean absolute error:\t"
+                      = cv.weighted_mae.round(3) if cv.weighted_mae
+                      %br 
+                      = "R square:\t"
+                      = cv.r_squared.round(3) if cv.r_squared
+                      /%br
+                      /= "Correlation plot"
+                      /= cv.correlation_plot
+                      /%br
+                      /= "Confidence plot:"
+                      /= cv.confidence_plot
 
   
   %fieldset#bottom.well
diff --git a/views/prediction.haml b/views/prediction.haml
index e7bc88a..47bd9b7 100644
--- a/views/prediction.haml
+++ b/views/prediction.haml
@@ -22,6 +22,7 @@
             %p= @compound.svg
             %p= @compound.smiles
           - @predictions.each_with_index do |prediction,i|
+            - type = @models[i].model.class.to_s.match("Classification") ? "Classification" : "Regression"
             %td{:style=>"vertical-align:top;"}
               %b{:class => "title"}
                 = "#{@models[i].endpoint.gsub('_', ' ')} (#{@models[i].species})"
@@ -30,23 +31,23 @@
                 %p
                 / TODO fix scientific notation from database
                 %b Measured activity:
-                = prediction[:value].numeric? ? "#{prediction[:value].to_f.round(3)} (#{@models[i].unit})" : prediction[:value]
+                = (type == "Regression") ? "#{prediction[:value].round(2)} (#{@models[i].unit})" : prediction[:value]
                 %p Compound is part of the training dataset
               - elsif prediction[:neighbors].size > 0
                 %p
                 / model type (classification|regression)
                 %b Type:
-                = @models[i].model.class.to_s.match("Classification") ? "Classification" : "Regression"
+                = type
                 %br
                 %b Prediction:
                 / TODO scientific notation
-                = prediction[:value].numeric? ? "#{'%.2e' % prediction[:value]} #{@models[i].unit}" : prediction[:value]
+                = (type == "Regression") ? "#{'%.2e' % prediction[:value]} #{@models[i].unit}" : prediction[:value]
                 / TODO update description
                 / %a.btn.glyphicon.glyphicon-info-sign{:href=>"#", :title=>"Prediction", data: {toggle:"popover", placement:"left", html:"true", content:"LAZAR calculates searches the training dataset for similar compounds (neighbors) and calculates the prediction from their measured activities. LAZAR calculates predictions using <ul><li>a majority vote (weighted by compound similarity) for<br /><b>classification</b> (<a href='http://www.frontiersin.org/Journal/10.3389/fphar.2013.00038/abstract', target='_blank'>original publication</a>) </li><li>a local QSAR model based on neighbors for<br /><b>regression</b> (<a href='http://www.frontiersin.org/Journal/10.3389/fphar.2013.00038/abstract', target='_blank'</h>original publication</a>) </li></ul>Please keep in mind that predictions are based on the measured activities of neighbors."}}
                 %br
                 / TODO probability
                 %b Confidence:
-                = prediction[:confidence].round(3)
+                = prediction[:confidence].round(2)
                 / %a.btn.glyphicon.glyphicon-info-sign{:href=>"#", :title=>"Confidence", data: {toggle:"popover", placement:"left", html:"true", content:"Indicates the applicability domain of a model. Predictions with a high confidence can be expected to be more reliable than predictions with low confidence. Confidence values may take any value between 0 and 1. For most models confidence > 0.025 is a sensible (hard) cutoff to distinguish between reliable and unreliable predictions."}}
                 %p
                 /TODO add tooltip for significant ftagments and descriptors