launch_a_data_science_environment_in_oracle_cloud.pdf
- 1. Launching a Data Science Environment on Oracle Cloud Infrastructure By Nicholas Toscano
- 2. 1. Open the Resource Manager 2. Create a Data Science Stack 3. Create a Data Science Project 4. Create a Notebook Session 5. Start a Notebook Session 6. Install a Pre-Configured Conda Pack 7. Launch an Example Notebook Steps to Launch a Data Science Environment
- 3. Open the OCI Resource Manager Fast deploy a data science environment via the Resource Manager. Use a pre-created stack.
- 4. Create a Data Science Stack Select the Data Science Service. This will provision Data Science and its prerequisites. Go ahead and select the option to create one or more notebooks. Use an E2.8 or a GPU2.1
- 5. Create additional notebooks as required. 1 notebook can hold multiple data science projects, environments and kernels. You can store some data in the notebook (not recommended) or call data from REST. Set up ADW storage. Create a Data Science Project
- 6. Create a Notebook Session Launching through Resource Manager will provide you with options for conda packs and notebook examples. Remember this is a Python 3 environment. Launching conda packs will give you alternative kernels.
- 7. Install a Pre-Configured Conda Pack Environment Explorer contains preconfigured conda packs. Start with: General Machine Learning for CPUS/GPUS Run in terminal: CPU: $ odsc conda install -s mlcpuv1 GPU: $ odsc conda install -s mlgpuv1
- 8. Review Pre-installed Libraries Use âpip freezeâ, âconda listâ, or âconda env ân baseâ list to view the managed libraries in base environment. You can run an update to make sure the versions are complete. You can install libraries that are not present.
- 9. Launch an Example Notebook Start with Binary Classification. Go over ADS. You can retain or save your notebooks in the File Browser. SHUTDOWN YOUR SERVICES WHEN YOU DONE
- 10. END