Project 2 Summary

Overall the class did a great job!

14/29 were 19 or higher!

  • 19+ (14 projects): 20, 20, 20, 20, 20, 20, 20, 19.5, 19.5, 19.5, 19.5, 19.5, 19.25, 19

  • 18+ (3 projects): 18.5, 18.25, 18

General Comments: Kudos and Cautions

The scores were very good, and, generally speaking, people are understanding the material. We want to mention a few issues that came up.

Kudos

  1. Several people used data standardization and hyperparameter tuning techniques.

  2. There were a few who used pipelines, which was impressive.

  3. Several people experimented with other techniques, such as Histogram-based gradient boosting.

Cautions

  1. Many people lost points for not commenting on data anomalies derived from the statistical information. Additionally, some did not provide insights from the univariate analysis.

  2. Some people did not realize that they needed to perform hyperparameter tuning. Others performed hyperparameter tuning but used fewer than two parameters.

  3. A few people did not apply any data standardization techniques or performed incorrect standardization. Note: Data standardization is always done after splitting the data into training and test sets; otherwise, it will cause data leakage.

  4. Very important Please ensure that your notebook and report are consistent in terms of techniques and observations. We were lenient this time and awarded partial credit, but next time, we plan to deduct full points for inconsistencies.

  5. Lastly, please consider splitting code into multiple cells and adding comments. This will improve code readability.