nestauk / dap_taltech

Tutorials for taltech hack week 2023
MIT License
2 stars 1 forks source link

Time Series #21

Closed india-kerle closed 10 months ago

india-kerle commented 10 months ago

This PR:

  1. Adds the notebooks necessary for the time series tutorial;
  2. Adds horizon 2020 getters in the data_getters.py notebook;
  3. Adds plotting functionality to utils.

To review the PR, work through the notebooks in the time_series directory.

ampudia19 commented 10 months ago

Review of the Notebook

The notebook is well-structured, and I never get tired of the emojis and suggested tasks. Great stuff. Minor changes, already implemented and pushed.

A few comments

Data

  1. Missing Values: You checked for missing values but didn't discuss what you would do if you found any. It might be helpful to add a comment or two about potential strategies for handling missing data.

  2. Correlation Matrix: While you've plotted a correlation matrix, it would be beneficial to discuss the implications of the correlations observed, especially multicollinearity.

Hyperparameter Tuning

  1. Grid Search vs Random Search: You've used both methods but haven't discussed the trade-offs between them. It may also be useful to drop "Bayesian Optimisation" as a third family that we ignore here.