The PySD Cookbook

Simple recipes for powerful analysis of system dynamics models

Quickstart

1. Install PySD (we recommend using the latest available PySD version) and needed requirements for the notebooks

2. Download this cookbook and unzip it to a working directory

3. Navigate your command prompt to the working directory just created

4. Launch ipython notebook with the command ipython notebook at your command prompt

5. In the ipython browser, open sourceanalysesgetting_startedHello_World_Teacup.ipynb

How to use this cookbook

Every recipe in this cookbook is an executable ipython notebook. Because of this, I recommend that you download a copy of the cookbook, and follow along by executing the steps in each notebook as you read, playing with the parameters, etc.

If you want to implement a recipe, you can then make a copy of the notebook you are interested in, and modify it to analyze your own problem with your own data.

To download the cookbook in its entirity, use this link or visit the cookbook’s Github Page and select one of the options in the righthand panel.

Contents

• Getting Started
  • Introduction to the PySD Cookbook
  • Installation and Setup of Python and PySD
  • Getting Started with Python
  • Using the Jupyter Notebook
  • Working with git and github
  • Hello World: The Teacup Model
  • SIR Peak Infection Challenge
• Data Handling
  • Data handling with Pandas
  • Saving Simulation Results to a Database
  • Exogenous model input from a file
• Desing Policy
  • PySD EMAWorkbench Connector Demo
• Visualization
  • Basic Visualization with matplotlib
  • Other Visualization Packages in Python
  • Phase Portraits
  • Demo of a KDE plot beside timeseries set
  • Demo of a KDE plot beside timeseries set (Interactive)
  • Plotting a Suite of Simulations
• Model Fitting
  • Fitting a model’s parameters with run-at-a-time optimization
  • Step-at-a-time optimization
  • Parallel Model Fitting
  • Fitting a model with Markov Chain Monte Carlo
  • Estimating penny population parameters
• Geographic Analyses
  • Doing more with spatial data
  • Using SD models to understand the differences between population measures at varying levels of geographic disagregation
• Model Comparison
  • Model Assessment
  • Designing Experiments
• Sensitivity
  • Latin Hyopercube Sampling
• Surrogating Functions
  • Surrogating a function with a machine learning estimator
• Testing
  • Formal Behavior Testing
• Realtime Data Incorporation
  • Ingredients
  • Feeding models with realtime streaming data
• Model Development Workflow
  • Unit Testing
• Wrapper EMAWorkbench
  • Notebook illustrating using PYSD and EMA
  • Example Simple SIR Model
• Data Used in this Cookbook
  • Baby Name Data
  • Collecting US decennial census data
  • Carbon Data Sources
  • Manufacturing Defects Synthetic Data
  • Ebola Data Loader
  • Parse gmail .mbox file
  • Town Formatter
  • Prepare georeferenced data for artificial disease outbreak in Europe
• Chapters to be Written
• End Notes
  • For instruction

Extra Resources

The PySD Documentation can answer most questions about how PySD itself should be used, with basic examples and function reference.

Indices and tables

• Index

• Module Index

• Search Page