Update documentation and fixed tests

README.md

@@ -10,33 +10,33 @@ For more details, see full paper [DeepSurv: Personalized Treatment Recommender S
 
 ## Installation:
 
-### Dependencies
+### From source
 
-Theano, Lasagne (bleeding edge version), lifelines, matplotlib (for visualization) and all of their respective dependencies. 
+Download a local copy of DeepSurv and install from the directory:
 
-### Installing
+	git clone https://github.com/jaredleekatzman/DeepSurv.git
+	cd DeepSurv
+	pip install .
 
-You can install *DeepSurv* using
+### Dependencies
 
-	pip install deepsurv
+Theano, Lasagne (bleeding edge version), lifelines, matplotlib (for visualization), tensorboard_logger, and all of their respective dependencies. 
 
-from the command line.
+### Running the tests
 
-#### Dependencies
+After installing, you can optionally run the test suite with
 
-DeepSurv has been succesfully installed and tested on macOS Sierra 10.12.1 with the follow versions of Python packages:
+	py.test
 
-        'theano==0.8.2',
-        'lasagne==0.2.dev1',
-        'lifelines==0.9.2',
+from the command line while in the repo's main directory. 
 
-### Running the tests
+## Running Experiments
 
-After installing, you can optionally run the test suite with
+Experiments are run using Docker containers built off of the [floydhub](https://github.com/floydhub/dl-docker) deep learning Docker images. DeepSurv can be run on either the CPU or the GPU with nvidia-docker. 
 
-	py.test
+All experiments are in the `DeepSurv/experiments/` directory. 
 
-from the command line while in the module main directory.
+To run an experiment, define the experiment name as an environmental variable `EXPRIMENT`and run the docker-compose file. Further details are in the `DeepSurv/experiments/` directory. 
 
 ## Training a Network
 
@@ -62,10 +62,3 @@ You can then evaluate its success on testing data:
 If you have matplotlib installed, you can visualize the training and validation curves after training the network:
 
 	deepsurv.plot_log(log)
-
-## Running Experiments
-
-Experiments are run using Docker containers built off of the [floydhub](https://github.com/floydhub/dl-docker) deep learning Docker images. DeepSurv can be run on either the CPU or the GPU with nvidia-docker. 
-
-To run an experiment, define the experiment name as an environmental variable `EXPRIMENT`and run the docker-compose file. 	
-

experiments/README.md

@@ -0,0 +1,37 @@
+# DeepSurv Experiments
+
+Experiments are run using Docker containers built off of the [floydhub](https://github.com/floydhub/dl-docker) deep learning Docker images. 
+
+### Requirements
+
+Download and install [Docker](https://www.docker.com/community-edition#/download). If you plan to use FloydHub's GPU tag, install [nvidia-docker](https://github.com/NVIDIA/nvidia-docker).
+
+### Experiments
+
+To run one of the experiments from the paper, use the following command from this directory:
+
+	 EXPERIMENT=${EXPERIMENT_ID} docker-compose up --build
+
+The following experiments are provided:
+
+| Experiment    | `${EXPERIMENT_ID} `|
+| ------------- |:-------------:|
+| Simulated Linear Data | `linear` |
+| Simulated Nonlinear (Gaussian) Data | `gaussian` |
+| Worchester Heart Attack Study (WHAS) | `whas` |
+| Study to Understand Prognoses Preferences Outcomes and Risks of Treatment (SUPPORT) | `support` |
+| Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) | `metabric` |
+| Simulated Treatment Data | `treatment` |
+| Rotterdam & German Breast Cancer Study Group (GBSG) | `gbsg` |
+
+The hyper-parameters for each experiment are in this repo's directory: `DeepSurv/experiments/deepsurv/models/`
+
+For more details on each experiment, reference [DeepSurv: Personalized Treatment Recommender System Using A Cox Proportional Hazards Deep Neural Network](http://arxiv.org/abs/1606.00931).
+
+#### Using GPU
+
+If you have nvidia-docker installed, you can run DeepSurv experiments using your GPU. To do so change the tag in the first line of your experiment's docker file. 
+
+For example, to run the simulated linear data experiment with the GPU change the first line the file `./deepsurv/Docker.linear` to:
+
+	FROM floydhub/dl-docker:gpu	

requirements.txt

@@ -2,5 +2,5 @@ h5py==2.7.0
 lifelines==0.9.4
 logger==1.4
 Optunity==1.1.1
-tensorboard-logger==0.0.3
+tensorboard_logger==0.0.3
 lasagne==0.2.dev1

setup.py

@@ -4,7 +4,7 @@
     long_description = file.read()
 
 setup(name='deepsurv',
-    version='0.1.0',
+    version='0.2.0',
     description='Deep Cox Proportional Hazards Network for deep learning and survival analysis.',
     keywords = "survival analysis deep learning cox regression",
     url='https://github.com/jaredleekatzman/DeepSurv',
@@ -24,6 +24,7 @@
         'theano==0.8.2',
         'lasagne==0.2.dev1',
         'lifelines==0.9.2',
+        'tensorboard_logger==0.0.3'
     ],
     # test_suite = 'nose.collector',
     # test_require = ['nose','lasagne','theano']

tests/test_deepsurv.py

@@ -43,10 +43,6 @@ def test_deepsurv_initialize_dropout(self):
         network = DeepSurv(dropout = 0.5, **self.hyperparams)
         assert len(network.hidden_layers) == 2 * len(self.hidden_layers_sizes) + 1
 
-    def test_deepsurv_initialize_standardize_layer(self):
-        network = DeepSurv(standardize = True, **self.hyperparams)
-        assert len(network.hidden_layers) == len(self.hidden_layers_sizes) + 3
-
 class TestDeepSurvTrain():
 
     @classmethod
@@ -65,12 +61,10 @@ def setup_class(self):
         self.network = network
 
     def test_train(self):
-        # Test if network computes nan for any values
-        assert self.log.has_key('train') == True
-        assert numpy.any(numpy.isnan(self.log['train'])) == False
-
-        assert self.log.has_key('valid') == True
-        assert numpy.any(numpy.isnan(self.log['valid'])) == False
+        # Test if network has undefined parameters
+        assert self.log.has_key('best_params') == True
+        params_is_nan = [is_nan for params in self.log['best_params'] for is_nan in numpy.isnan(params.flatten())]
+        assert numpy.any(params_is_nan) == False
 
     def test_network_predict_risk(self):
         risk = self.network.predict_risk(self.test['x'])