Improved the TFX OSS notebook and README

Addressed the ussues discovered here: kubeflow#913

samples/tfx-oss/README.md

@@ -9,6 +9,8 @@ This pipeline demonstrates the TFX capablities at scale. The pipeline uses a pub
 
 ## Setup
 
+Enable DataFlow API for your GKE cluster: <https://console.developers.google.com/apis/api/dataflow.googleapis.com/overview>
+
 Create a local Python 3.5 conda environment
 ```
 conda create -n tfx-kfp pip python=3.5.3
@@ -29,7 +31,13 @@ git clone https://github.com/tensorflow/tfx
 
 Upload the utility code to your storage bucket. You can modify this code if needed for a different dataset.
 ```
-gsutil cp tfx/examples/chicago_taxi_pipeline/taxi_utils.py gs://my-bucket/
+gsutil cp tfx/examples/chicago_taxi_pipeline/taxi_utils.py gs://my-bucket/<relative path>/
+```
+
+If gsutil does not work, try `tensorflow.gfile`:
+```
+from tensorflow import gfile
+gfile.Copy('tfx/examples/chicago_taxi_pipeline/taxi_utils.py', 'gs://<my bucket>/<relative path>/taxi_utils.py')
 ```
 
 ## Configure the TFX Pipeline
@@ -39,9 +47,9 @@ Modify the pipeline configuration file at
 tfx/examples/chicago_taxi_pipeline/taxi_pipeline_kubeflow_large.py
 ```
 Configure 
-- GCS storage bucket name (replace "my-bucket")
-- GCP project ID (replace "my-gcp-project")
-- Make sure the path to the taxi_utils.py is correct
+- Set `_input_bucket` to the GCS directory where you've copied taxi_utils.py. I.e. gs://<my bucket>/<relative path>/
+- Set `_output_bucket` to the GCS directory where you've want the results to be written
+- Set GCP project ID (replace my-gcp-project). Note that it should be project ID (usually has numbers in the end), not project name.
 - Set the limit on the BigQuery query. The original dataset has 100M rows, which can take time to process. Set it to 20000 to run an sample test.
 
 

samples/tfx-oss/TFX Example.ipynb

@@ -19,6 +19,14 @@
     "!pip3 install https://storage.googleapis.com/ml-pipeline/release/0.1.10/kfp.tar.gz --upgrade\n"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Enable DataFlow API for your GKE cluster\n"
+    "<https://console.developers.google.com/apis/api/dataflow.googleapis.com/overview>\n"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -41,8 +49,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# copy the trainer code to a storage bucket \n",
-    "!gsutil cp tfx/examples/chicago_taxi_pipeline/taxi_utils.py gs://<my bucket>/"
+    "# copy the trainer code to a storage bucket as the TFX pipeline will need that code file in GCS\n",
+    "from tensorflow import gfile\n",
+    "gfile.Copy('tfx/examples/chicago_taxi_pipeline/taxi_utils.py', 'gs://<my bucket>/<relative path>/taxi_utils.py')"
    ]
   },
   {
@@ -57,9 +66,9 @@
     "```\n",
     "\n",
     "Configure:\n",
-    "- GCS storage bucket name (replace my-bucket)\n",
-    "- GCP project ID (replace my-gcp-project)\n",
-    "- Make sure the path to the taxi_utils.py is correct\n",
+    "- Set `_input_bucket` to the GCS directory where you've copied taxi_utils.py. I.e. gs://<my bucket>/<relative path>/\n",
+    "- Set `_output_bucket` to the GCS directory where you've want the results to be written\n",
+    "- Set GCP project ID (replace my-gcp-project). Note that it should be project ID (usually has numbers in the end), not project name.\n",
     "\n",
     "The dataset in BigQuery has 100M rows, you can change the query parameters in WHERE clause to limit the number of rows used.\n"
    ]
@@ -89,12 +98,17 @@
     "# Get or create a new experiment\n",
     "import kfp\n",
     "client = kfp.Client()\n",
-    "experiment = client.create_experiment(\"TFX Examples\")\n",
-    "pipeline_filename = \"chicago_taxi_pipeline_kubeflow.tar.gz\"\n",
+    "experiment_name=\"TFX Examples\"\n",
+    "try:\n",
+    "    experiment_id = client.get_experiment(experiment_name=experiment_name).id\n",
+    "except:\n",
+    "    experiment_id = client.create_experiment(experiment_name).id\n",
+    "\n",
+    "pipeline_filename = "chicago_taxi_pipeline_kubeflow.tar.gz"\n",
     "\n",
     "#Submit a pipeline run\n",
     "run_name = 'Run 1'\n",
-    "run_result = client.run_pipeline(experiment.id, run_name, pipeline_filename, {})\n"
+    "run_result = client.run_pipeline(experiment_id, run_name, pipeline_filename, {})\n"
    ]
   }
  ],