The purpose of this exercise is to get you focused on some key questions that may come up during an interview. 🚀
You may not know answers to the questions below, let alone how it works within code. That's okay! Do some research of your own and try to understand the concepts.
Copy the below questions into Microsoft Word, Google Docs, or duplicate the page into your own Notion page. Then under each question write out an answer to the question in 2 - 3 lines.
As you learn more technologies and tools, you can then add to the list below. Over time, you will then build a list of key concepts that you can explain.
- What are model bias and variance? What is bias-variance trade-off?
- What is data normalisation? Why is it particularly important prior to modelling a KNN algorithm? Name 2 normalisation techniques and explain how they work.
- Is Logistic Regression a regression or classification algorithm? What function is the backbone of the algorithm? What is the precision/recall tradeoff?
- Why do we use cross validation to score Machine Learning models?
- Describe the following metrics and give a use case for each: recall, precision, accuracy, r2.
- What is a decision tree, and how is it “grown”? What is the bootstrap aggregated version of decision trees, and what is its purpose?
- What is “boosting”, and what is its purpose? On which Machine Learning model can boosting be implemented?
- Name two techniques for optimizing a machine learning model’s hyperparameters. What are the differences between the two?
- What is dimensionality reduction and when would it be useful? Name a dimensionality reduction technique.
- In neural networks, what is transfer learning?
...Add your own questions below and keep a list of other interview questions