Read Me

Table of Contents

• Added files
• Installation
• Usage

Added files

The following 3 files have been added to the model provided and will be explained below:

1. analysis_transport.ipynb

This notebook uses certain data files to determine the economic added value of transport over the IJssel river. It requires only pandas to run and matplotlib to visualise. For this notebook to run, the following data files are required:

• data/economic_benefit.csv
• data/economic_value.csv
• data/revenue_transport.csv
• data/transport_from.csv
• data/transport_to.csv

2. dike_model_optimization.ipynb

This notebook is the crux of our analysis, and walks through all the steps of Multi-Scenario MORDM. These steps are outlined in the notebook and the results are analysed separately in the report. The file requires most of the requirements stated in installation. As some optimizations and experiments take a long time to run, pickle is used to dump output formats into the archives directory. This directory contains the following:

• step1_convergences: stores the list of convergence values of the first optimization
• step1_results: stores the list of results of the first optimization
• step3_results.tar.gz: stores the results of the experiments after the first optimization.
• step5_results: stores the results of the second optimization run.
• step7_results: stores the results of the experiments after the second optimization.

3. preliminary_sensitivity.py

The sensitivity analysis provided in the code uses several techniques to evaluate the sensitivity of the results provided by the dike model. It uses different functions who each represent their own method of analysis to assess the impact of uncertainties on key performance indicators (KPIs), all methods come together in a final function. Each of them provides a certain results each saved in a corresponding directory, either in the form of a text file, csv file or png file. The defined KPIs are expected number of deaths, investment costs of building dikes, invest costs for the room for the river, expected annual damages and evacuation costs

Two other files that are part of the model were used in the sensitivity analysis

1. from dike_model_simulation: The function 'problem_formulation'
2. problem_formulation: The function 'get_model_for_problem_formulation' They were used in order to receive the uncertainty parameters of the model

Installation

Besides the 3.11 version of python and the ema-workbench make sure the following packages are installed before trying to run the code:

1. pandas
2. Numpy
3. matplotlib
4. seaborn
5. statsmodels
6. SALib
7. scikit-learn

Usage

The notebooks themselves are self-explanatory, but for the sensitivity the following instructions are provided:

• The code consist out of several functions which are all called upon by one function: run_sensitivity_analysis(sobol, regular, name) This requires you to enter the directory path of the sobol run, the directory path of th regular run and the name under which you would like the run to be saved. After running the analysis this folder can be found inside the 'results' folder. It will produce image files of the graphs produced by the convergence, sobol and tree analysis.

  Examples

  An example of the lines to actually run the code is provided by the following: run_sensitivity_analysis("results/run_sobol_base.tar.gz", "results/run_base.tar.gz", 'Base_run_0_policies') run_sensitivity_analysis("results/run_sobol_policies.tar.gz", "results/run_policies.tar.gz", 'Base_run_10_policies') Here the analysis is run for both a zero policies run and a 10 policies run, it is an exploratory analysis.