Hello students! 👋 This Future Ready Skills Grab Challenge provides a self-paced learning path for data science and machine learning. After completing foundational content on Microsoft Learn, a challenge using Grab data will be available for you to test your skills.
Ideally, students will have had some exposure to data science and machine learning already.
Choose one of the following Future Ready Skills Challenges:
See below for suggestions on which technologies to use. The most important that we can review the code on Github.
Feel free to use whatever system or technology you are most comfortable with (bonus points for using Microsoft services). We would highly encourage you to use some form of Jupyter Notebook, so that you can easily comment your progress and visualize intermediate results. Here are a couple of suggestions:
Visual Studio Code is Microsoft's multi-platform, lightweight, free and open-source code editor. It comes with Github integration and also Jupyter Notebook integration. Follow these links for more information. Choosing VS Code allows you to download and analyze the data on your own computer.
The data science VM is a server (VM) running on Azure that is preloaded with a variety of ML software and tools. You can run the DSVM as Windows or Linux and access it via RDP (Windows) or SSH (Linux). Alternatively, you can access its Jupyter Notebook server just using your browser.
To start such a machine, you will need an Azure account. The easiest is to sign up for a free Azure for Students account, which gives you $100 (yearly renewing) credit. Then have a look at this intro to Azure VM to get started.
Also have a look at this introduction to Data Science on Azure or this article on how to choose the Data Science service in Azure you need (45 min)
Another alternative is to use Azure Machine Learning services. This basically comes in two flavours: the graphical and free Machine Learning Studio or Azure ML. Machine Learning Studio is great to get started in a low-code environment. Azure ML is a enterprise grade environment, that has Jupyter Notebooks built-in, automatically logs all experiments and is backed by autoscaling compute clusters.
You will need an Azure account to be able to use Azure ML. See above for more info.
NOTE: Azure Notebooks will be retired in October 2020.
Azure Notebooks is a free and Microsoft hosted Jupyter Notebook service. It allows you to explore the data and write some code in your web browser, without having to install software locally. To track code changes over time we will use Github, which will also allow you to easily change systems later, if needed. Follow this link, to read how to set up Azure Notebooks and Github. Also have a look at this introduction to machine learning with Python and Azure Notebooks.
The students are expected to submit the following:
- Your script: Either make your Azure Notebook public and share the URL in the Microsoft Teams channel or add us as collaborator to your Github repository containing the solution.
- Either a report or ppt slides (less than 30 pages) contains but not limit to your business understandings, data insights findings, feature engineering ideas, model selection strategies, conclustions and recommendations, etc.
- Submit a presentation vedio (less than 20 min) introduce your project and models.
- Data Insights (20%)
- Feature Engineering (20%)
- ML Model (20%)
- Parameter Tuning (20%)
- Readability (20%)
If you would like to conduct self-paced training to have deeper understand on traning machine learning models, please refer to this link -> https://github.com/yingjia-liu/Msft-Grab-FRS/tree/master/ML-Model-Tutorials
Good Luck 😃