HPDM097: Making a difference with health data:

Forecasting health service demand

Forecasting practical materials for Making a difference with health data module.

Dependencies

Please use the provided conda environment

conda env create -f binder/environment.yml

conda activate hds_forecast

Citation:

Monks, T. (2023). forecasting health service demand in python. Zenodo. https://doi.org/10.5281/zenodo.4332600

@software{monks_2023_10370697,
  author       = {Monks, Thomas},
  title        = {forecasting health service demand in python},
  month        = dec,
  year         = 2023,
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.4332600},
  url          = {https://doi.org/10.5281/zenodo.4332600}
}

Syllabus

RECOMMENDED Pre-course material

These notebooks offer a refresher in the basics of date handling in numpy, pandas and matplotlib.

• Handling dates in numpy and pandas:

• Exploring time series with pandas and matplotlib:

Computer Lab 1: The basics of forecasting: part 1

1.1 Code along notebooks

These notebooks accompany the exercises. They provide example code to help you solve the exercises.

• Loading time series data into pandas:

• Naive benchmarks:

1.2 Exercises

• Naive forecasting exercises

Computer Lab 2: The basics of forecasting: part 2

2.1 Code along notebooks

• Introduction to cross validation

2.2 Exercises

• Time series cross validation

Computer Lab 3: Forecasting using ARIMA models

3.1 Code along notebooks

• Introduction to ARIMA Exercises:

3.2 Exercises

• ARIMA Exercise:

Computer Lab 4: Forecasting daily data using Facebook Prophet

4.1 Code along notebooks

• Prophet introductory lecture:

• Introduction to Prophet Exercises:

4.2 Exercises

• Prophet Exercises:

Computer Labs 5. An introduction to feedforward neural networks

5.1. code along lecture notebooks

• Deep Learning 101:

5.2 Exercises

• Autoregressive Neural Networks with KERAS. Part 1:

5.3. Optional self study material

• Preprocessing and autoregressive OLS:

• Autoregressive Neural Networks PYTORCH. Part 1:

Computer Lab 6: Feedforward neural networks for time series

6.1 Exercises

• Autoregressive Neural Networks KERAS Part 2:

6.2 Optional self study material

• Autoregressive Neural Networks PYTORCH Part 2: