There are few steps that needs to be performed to use Snap ML in DSXL environment. These are outlined as follows:

  1. Go to the IBM DSXL web console.
  2. Login with your username and password.
  3. On the IBM DSXL homepage dashboard, click Add project.
  4. Once you click Add project, a new window named Create Project appears where you can create a notebook from a file or a github repository. The window looks like the one shown below.

     

     

  5. In this example, we will use specific github repository links in order to use the snap-ml-local environment. For the current release, there are 2 such github repository links with example notebooks. These notebooks demonstrate the advantage of using snap-ml-local environment for a machine learning pipeline.
  6. Under New tab, type a name for the project and enter a brief description for the project. For example, create a project named snapml_dsxl as shown below.
     
     
  7. After the project is created, you will see a page like the one shown below.
     
     
  8. Under Assets, select notebooks and click Add Notebook. The Create Notebook window appears.
     
     
  9. If you want, you can set the number of CPU cores, the number of GPUs and memory limit depending upon the machine configuration.
     
     
  10. Under the project title snapml_dsxl, move to the From URL tab. Name the notebook. Add a description if you want. Add the notebook url as https://github.com/ibmsoe/snap-ml/blob/master/notebooks/credit-default-prediction-example.ipynb. Select the environment as Jupyter with Python 3.6 and PowerAI v1.5.3 for GPU. The window appears as follows after all these entries.
     
     

Finally, you will see the jupyter notebook window similar to the following. It shows an example application for the task of credit default risk prediction using SnapML in DSXL.
 
 

Join The Discussion

Your email address will not be published. Required fields are marked *