Assignment 1.3: AutoGluon Lab Questions

Run the AWS Academy Lab 3: Getting Started with AutoGluon.
1. All assets: mod01_lab03_autogluon.zip
  1. Note that autogluon does not support Python beyond 3.12
2. Colab currently has Python 3.12 -- https://colab.research.google.com/
  1. Code changes
    - Replace !pip install -U -q -r requirements.txt with !pip install autogluon
    - Delete the line %load_ext autoreload from the next cell
  2. Upload the data files (test.csv and train.csv) and the Python quiz file (MLUMLA_EN_M1_Lab3_quiz_questions.py), placing all three (3) files in the main folder:
Answer the questions at the end, which are reproduced below--and also provide at least one screenshot showing that you've run the lab. (type text and paste screenshot(s) into a single Word doc):
1. What is the shape of the training dataset?
2. What type of ML problem (i.e., classification or regression) does AutoGluon infer? (Hint: Remember, you didn't mention the problem type. You only provided the label column.)
3. What does AutoGluon suggest in case it inferred the wrong problem type?
4. What kind of data preprocessing and feature engineering did AutoGluon perform?
5. What are the basic statistics about the label in the print statements from AutoGluon?
6. How many extra features were generated in addition to the originals in the dataset? What was the runtime for that?
7. Which evaluation metric was used?
8. What does AutoGluon suggest in case it inferred the wrong metric?
9. What is the ratio between the training and validation dataset? (Hint: Look for val or validation.)
10. Where did AutoGluon save the predictor?
11. Which folder were the models saved in?
12. What file format are the models in? (Note: Look at the file name suffix. You don't need to open the file.)

SUBMIT in a single Word doc (docx):

numbered text of your responses to the above questions
at least one pasted screenshot showing you ran the lab