forecasting - v5.0.0

Changed

• ENV: Updated hds_forecast to python 3.11 and latest compatible packages July 2024 (note tensorflow set to 2.15 to match Google Colab).
• ENV: Now uses forecast-tools v0.3.0

Fixed

• NN: sliding_window patch to include final window (loop + 1) #30
• NN: Notebooks patched to include code to check and then create output directory if it does not exist
• PRE-REQS: notebooks fixed to download data from Github URL. #6
• LAB2: Exercise 1.4 fixed to use correct prediction horizon #24
• WEEK1 EXTRA: Patched to download dat from Github URL #16

- Jupyter Notebook
Published by TomMonks 12 months ago

forecasting - v4.0.1

Changes:

• PATCH: hds_pytorch updated to use forecast-tools version 0.2.1

- Jupyter Notebook
Published by TomMonks about 2 years ago

forecasting - v4.0.0

Changes

• ENV: Updated hds_forecast to python 3.10 and latest package Dec 2024
• ENV: notebooks updated to work woth forecast-tools v0.2.1
• ENV: created separate hds_pytorch virtual environment to support pytorch notebooks in python 3.8 and pytorch 1.4
• ENHANCEMENT: added coverage metrics notebooks to lab 1 additional materials
• PATCH: minor updates fixing notebooks to work with data stored on github.
• PATCH: minor updates to notebooks to account for updates to packages.

- Jupyter Notebook
Published by TomMonks about 2 years ago

forecasting - v3.0.2

Changes:

• Bug fix: read me file contained typo for opening Google Colab week 1 exercises.

- Jupyter Notebook
Published by TomMonks about 3 years ago

forecasting - v3.0.1

Patch details:

• Renamed conda environment: hdsforecast2023 -> hds_forecast

- Jupyter Notebook
Published by TomMonks about 3 years ago

forecasting - v3.0.0

Changes

• Minor updates for the 2022/23 academics year
• Updated conda environment
• Minor changes to prophet and ARIMA code to account for updates in libaries

- Jupyter Notebook
Published by TomMonks about 3 years ago

forecasting - v2.2.0

Additions

Regression with ARIMA errors example notebook

Patches

keras autoregression_part.ipynb now works with sklearn v1.0.1

- Jupyter Notebook
Published by TomMonks about 4 years ago

forecasting - v2.1.0

Changes

• Conda environment:
  • pmdarima updated to v1.8.4
  • sklearn update to v1.0.1
• Lab 4: ARIMA comparison exercise updated with more information about warnings from statsmodels
• Simple RNN: updated to account for deprecations of Adam lr parameter.

- Jupyter Notebook
Published by TomMonks about 4 years ago

forecasting - v2.0.2

Patches:

• Fixes to remove deprecation warnings from notebooks:
  • Neural network autoregression notebooks: compile method parameter lr -> learning_rate
  • Prophet code along lecture: ax.grid(b=True) -> ax.grid(visible=True)
• Updated environment name: hds_forecast_2022 -> hds_forecast

- Jupyter Notebook
Published by TomMonks about 4 years ago

forecasting - v2.0.0

2021/22 Academic Year

v2.0.0 Changes

• ENV: updated packages to latest as of Dec 2021
• Note in particular we now pip install tensorflow-cpu (2.7.0). Same as Colab at time of release.
• fbprophet deprecated since v0.7.1. New method for pip installing prophet v1.0.1. Added pystan as separate pip install
• Minor updates and PEP8 clean-ups of notebooks.

- Jupyter Notebook
Published by TomMonks about 4 years ago

forecasting - v1.0.0

2020/21 Academic Year version 1.0.0

- Jupyter Notebook
Published by TomMonks about 5 years ago