Updated to correct missing data bug

greenelab · Jul 26, 2016 · 5fa2861 · 5fa2861
1 parent 7a96f9d
commit 5fa2861
Show file tree

Hide file tree

Showing 34 changed files with 427 additions and 5,843 deletions.
diff --git a/Figures.ipynb b/Figures.ipynb
diff --git a/README.md b/README.md
@@ -7,9 +7,9 @@ Denoising Autoencoder for Phenotype Stratification (DAPS) is a semi-supervised
 technique for exploring phenotypes in the Electronic Health Record (EHR).
 
 Upon build, figures are regenerated and saved in:
-https://github.com/greenelab/DAPS/blob/master/shippable/buildoutput/Figures.html
+[Images](https://github.com/greenelab/DAPS/tree/master/images)
 
-![](<./images/cluster2.png>)
+![](<./images/cluster.png>)
 
 Controls and 2 artificial subtypes of cases were simulated from 2 different
 models. The labels are the number of hidden nodes in the trained DAs. Principal
@@ -138,13 +138,13 @@ We've included 3 ipython notebook files to help analyze the results.
 Selected Results
 ----------------
 
-![](<./images/fig2.png>)
+![](<./images/figure_3_patients_100.png>)![](<./images/figure_3_patients_200.png>)![](<./images/figure_3_patients_500.png>)![](<./images/figure_3_patients_1000.png>)![](<./images/figure_3_patients_2000.png>)
 
 Classification AUC in relation to the number of labeled patients under
 simulation model 1 (RF – Random Forest, NN – Nearest Neighbors, DA – 2-node DA +
 Random Forest, SVM – Support vector machine).
 
-![](<./images/fig5.png>)
+![](<./images/fig2.png>)
 
 Case vs. Control clustering via principal components analysis and t-distributed
 stochastic neighbor embedding throughout the training of the DA (raw input to

diff --git a/classes/__pycache__/__init__.cpython-34.pyc b/classes/__pycache__/__init__.cpython-34.pyc
diff --git a/classes/__pycache__/da.cpython-34.pyc b/classes/__pycache__/da.cpython-34.pyc
diff --git a/classes/__pycache__/full_rf.cpython-34.pyc b/classes/__pycache__/full_rf.cpython-34.pyc
diff --git a/classes/__pycache__/nearest_neighbors.cpython-34.pyc b/classes/__pycache__/nearest_neighbors.cpython-34.pyc
diff --git a/classes/__pycache__/rfc.cpython-34.pyc b/classes/__pycache__/rfc.cpython-34.pyc
diff --git a/classes/__pycache__/sim.cpython-34.pyc b/classes/__pycache__/sim.cpython-34.pyc
diff --git a/classes/__pycache__/svm.cpython-34.pyc b/classes/__pycache__/svm.cpython-34.pyc
diff --git a/classes/__pycache__/tree.cpython-34.pyc b/classes/__pycache__/tree.cpython-34.pyc
diff --git a/classify_patients.py b/classify_patients.py
@@ -13,6 +13,7 @@
 
 from sklearn.cross_validation import StratifiedKFold
 from sklearn.metrics import roc_auc_score
+from sklearn.preprocessing import Imputer
 
 import classes.rfc as rfc
 import classes.svm as svm
@@ -28,7 +29,7 @@ def run(run_name='test', patient_counts=[100, 200, 500],
     # loop through patient files
     np.random.seed(seed=123)
     random.seed(123)
-  
+
     overall_time = time.time()
     i = 0
     path = './data/' + run_name + '/'
@@ -43,13 +44,18 @@ def run(run_name='test', patient_counts=[100, 200, 500],
                 patients = pkl.load(open(patients_path + '/' + file, 'rb'))
 
                 np.random.shuffle(patients)
+                X = patients[:, :-1]
+                y = patients[:, -1]
+
                 if d > 0:
-                    missing_vector = add_missing(patients, d)
+                    missing_vector = np.asarray(add_missing(patients, d))
+                    X = np.array(X)
+                    X[np.where(missing_vector == 0)] = 'NaN'
+                    imp = Imputer(strategy='mean', axis=0)
+                    X = imp.fit_transform(X)
                 else:
                     missing_vector = None
 
-                X = patients[:, :-1]
-                y = patients[:, -1]
                 print(sum(y), len(y))
 
                 dAs = {}

diff --git a/images/cluster2.png b/images/cluster2.png
diff --git a/images/fig3.png b/images/fig3.png
diff --git a/images/fig4.png b/images/fig4.png
diff --git a/images/fig5.png b/images/fig5.png
diff --git a/images/supp_figure_4_a_0.0.png b/images/supp_figure_4_a_0.0.png
diff --git a/images/supp_figure_4_a_0.1.png b/images/supp_figure_4_a_0.1.png
diff --git a/images/supp_figure_4_a_0.2.png b/images/supp_figure_4_a_0.2.png
diff --git a/images/supp_figure_4_a_0.3.png b/images/supp_figure_4_a_0.3.png
diff --git a/images/supp_figure_4_a_0.4.png b/images/supp_figure_4_a_0.4.png
diff --git a/images/supp_figure_4_a_0.9.png b/images/supp_figure_4_a_0.9.png
diff --git a/images/supp_figure_4_b_0.0.png b/images/supp_figure_4_b_0.0.png
diff --git a/images/supp_figure_4_b_0.1.png b/images/supp_figure_4_b_0.1.png
diff --git a/images/supp_figure_4_b_0.2.png b/images/supp_figure_4_b_0.2.png
diff --git a/images/supp_figure_4_b_0.3.png b/images/supp_figure_4_b_0.3.png
diff --git a/images/supp_figure_4_b_0.4.png b/images/supp_figure_4_b_0.4.png
diff --git a/images/supp_figure_4_b_0.9.png b/images/supp_figure_4_b_0.9.png
diff --git a/shippable.yml b/shippable.yml
@@ -22,7 +22,7 @@ build:
     - cd /root/src/github.com/greenelab/DAPS
     - nose2 --plugin nose2.plugins.junitxml --junit-xml test 
     - mv nose2-junit.xml shippable/testresults/tests.xml
-    - curl https://zenodo.org/record/46607/files/data.zip -o data.zip
+    - curl https://zenodo.org/record/58627/files/data.zip -o data.zip
     - unzip data.zip -d .
     - jupyter nbconvert --to html --execute ./Figures.ipynb
     - cp Figures.html shippable/buildoutput/

diff --git a/visualize/.ipynb_checkpoints/Cluster-checkpoint.ipynb b/visualize/.ipynb_checkpoints/Cluster-checkpoint.ipynb
diff --git a/visualize/.ipynb_checkpoints/VisSweep-checkpoint.ipynb b/visualize/.ipynb_checkpoints/VisSweep-checkpoint.ipynb