This dashboard provides a way for interested individuals to explore data regarding press and news related to CIERA.
Instructions are provided below for various levels of usage. Even if you have never edited code before, the goal of the instructions in Level 2 is for you to run the dashboard on your computer. On the other end of things, if you are comfortable with routine use of git, code testing, etc., then jump to Level 4 to get an overview of how the dashboard works and what you might want to edit.
- Level 0: Using the Dashboard Online
- Level 1: Changing the Configuration and Data
- Level 2: Using the Dashboard on your Computer
- Level 3: Making Some Edits to the Code
- Level 4: Significant Customization and Editing
- Level 5: Additional Features
The dashboard has a plethora of features that can be interacted with via a web interface. If the dashboard is currently live at ciera-press-zach, you can use the dashboard without any additional effort. One of the main features is the application of filters and the ability to download the edited data and images.
For many applications the data is light-weight enough that you can do the pre-processing as part of running the dashboard. If the dashboard is also hosted on the web you can edit the configuration and data without ever needing to download anything.
Some options are only available in the config.yml file found in the src directory (./src/config.yml if you are in the root directory, i.e. here).
You can edit this on github by clicking on the edit button in the upper right, provided you are logged in with an account that has the necessary permissions.
Locally this can be edited with TextEdit (mac), Notepad (Windows), or your favorite code editor.
The raw data lives in the data/raw_data folder.
To update the data used, add and/or replace the data in this folder.
You can do this on github by clicking the "Add file" button in the upper right hand corner.
The pipeline will automatically select the most recent data.
If you need a private dashboard or you need to run more-intensive data processing you'll need to run the dashboard on your computer.
The code lives in a git repository, but you don't have to know git to retrieve and use it. The process for downloading the code is as follows:
- Click on the green "Code" button on the GitHub repository, near the top of the page.
- Select "Download ZIP."
- Extract the downloaded ZIP file.
- Optional: Move the extracted folder (
press-dash; referred to as the code's "root directory") to a more-permanent location.
Running the dashboard requires Python. If you do not have Python on your computer it is recommended you download and install Miniconda.
With Python installed, open the directory containing the code (the root directory) in your terminal or command prompt. If youre a mac user and you've never used a terminal or command prompt before you can do this by right clicking the extracted folder and selecting "New Terminal at Folder" (more info; Windows Terminal is the windows equivalent).
Once inside the root directory and in a terminal, you can install the code by executing the command
pip install -e .
Inside the root directory and in a terminal window, enter
streamlit run src/dashboard.py
This will open the dashboard in a tab in your default browser. This does not require internet access.
To run the data-processing pipeline, while in the root directory run the following command in your terminal:
./src/pipeline.sh ./src/config.yml
Usage logs are automatically output to the logs directory.
You can open the notebooks as you would a normal Python notebook, if you are familiar with those.
A basic familiarity with git is highly recommended if you intend to edit the code yourself. There are many good tutorials available (e.g. GitHub's "Git Handbook", Atlassian Git Tutorial, Git - The Simple Guide, Git Basics). For convenience, the main command you need to download the code with git is
git clone git@github.com:CIERA-Northwestern/press-dash.git`
If you want to change how the data is processed, edit src/transform.ipynb.
The data-processing pipeline runs this notebook when you execute the bash script ./src/pipeline.sh,
and saves the output in the logs.
It is recommended to use the config whenever possible for any new variables introduced.
You can add additional notebooks to the data-processing pipeline.
Just make the notebook, place it in the src dir, and add its name to the array at the top of src/pipeline.sh.
The interactive dashboard is powered by Streamlit, a Python library that enables easy interactive access. Streamlit is built on a very simple idea---to make something interactive, just rerun the script every time the user makes a change. This enables editing the streamlit script to be almost exactly like an ordinary Python script. If you know how to make plots in Python, then you know how to make interactive plots with Streamlit.
If you want to change the Streamlit dashboard, edit src/dashboard.py.
Much of the Streamlit functionality is also encapsulated in utility functions inside the press_dash_lib/ directory, particularly in press_dash_lib/streamlit_utils.py.
Streamlit speeds up calculations by caching calls to functions.
If a particular combination of arguments has been passed to the function
(and the function is wrapped in the decorator st.cache_data or st.cache_resource)
then the results are stored in memory for easy access if the same arguments are passed again.
Before making significant edits it is recommended you make your own fork of the dashboard repository, and make your own edits as a branch. This will enable you to share your edits as a pull request.
The repository is structured as follows:
press-dash/
│
├── README.md # Documentation for the project
├── __init__.py
├── src # Source code directory
│ ├── __init__.py
| ├── config.yml # Configuration file for the dashboard
│ ├── dashboard.py # Script for interactive dashboard
│ ├── pipeline.sh # Shell script for running data pipeline
│ └── transform.ipynb # Jupyter notebook for data transformation
├── root_dash_lib # Custom library directory
│ ├── __init__.py
│ └── streamlit_utils.py # Utility functions for the dashboard
├── setup.py # Script for packaging the project
├── requirements.txt # List of project dependencies
├── data # Data storage directory
│ ├── raw_data # Raw data directory
│ └── processed_data # Processed data directory
├── test # Test directory
│ ├── __init__.py
│ ├── test_pipeline.py # Unit tests for data pipeline
│ └── test_streamlit.py # Unit tests for the dashboard
├── conftest.py # Configuration for test suite
└── test_data # Test datasets
The dashboard comes with a suite of code tests that help ensure base functionality. It is recommended you run these tests both before and after editing the code. To run the tests, simply navigate to the code's root directory and enter
pytest
If your edits include new packages, you need to add them to both requirements.txt and setup.py.
You may also consider changing the metadata in setup.py.
You can deploy your app on the web using Streamlit sharing. Visit Streamlit Sharing for more information.
Continuous integration (automated testing) is an excellent way to check if your dashboard is likely to function for other users. You can enable continuous integration via GitHub Actions, including adding a badge showing the status of your tests. Continuous integration can be tested locally using act. Some tests don't work on continuous integration, and are disabled until the underlying issues are addressed.
Streamlit has the option to deploy your code without sharing it publicly. More information can be found in this section of the Streamlit Sharing documentation.
ChatGPT was used in the construction of this document.