/rapid

Data-connected jupyter notebook for the RAPID paper

Primary LanguageJavaScriptMIT LicenseMIT

Rapid

This repo contains code, data and jupyter notebook related to RAPID.

Click this button to access the jupyter notebook without downloading/installing Binder

If the above link does not launch. Try an alternative server: Binder

Repository contents

The following sections indicate the folders which contain code and related data

Jupyter notebooks

  1. RAPID.ipynb - Notebook containing all visualizations
  2. Visualization and Statistical Analysis.ipynb - Initial analysis notebook
  3. Perovskite learning curve.ipynb - ML code and visualizations notebook

Raw data

All raw data files are located in the data folder

  1. cifs - Contains the Crystallographic Information Files
  2. images - Contains side vial images of each experiment performed
  3. xrd/xy - Contains xy files for XRD data
  4. 0042.perovskitedata_RAPID.csv - Escalate generated data file including 8 experimental features (with "rxn" as header prefix) and 67 chemical features (with "feat" as header prefix). The detailed explanations of these features are listed in "Explanation of Features-Descriptors" section in "Perovskite Dataset Description.pdf". This CSV file is used in visualization and machine learning.
  5. 0042.perovskitedata_RAPID_full.csv - This escalate generated data file contains the same experiments as "0042.perovskitedata_RAPID.csv" but has all 787 features, including additional "raw" features describing experiment details (see "Explanation of Features-Descriptors" section in "Perovskite Dataset Description" for the explanations of "raw" prefix). The csv file is not used for visualization or machine learning.
  6. image_list.json - Keeps track of all image files in the image folder
  7. ml_data.pkl - Python pickle file containing ML results
  8. inventory.csv - Chemical inventory data
  9. organic_inchikey.csv - Inchi keys and chemical names
  10. s_spaces.json - Co-ordinates of state space for each amine

Scripts

The following python scripts are used in the RAPID.ipynb notebook to generate visualizations

  1. plots.py - Generates the reaction outcomes 3D plot widget
  2. xrd_plot.py - Generates the xrd plot widget
  3. ml_section.py - Generates machine learning outcomes widget
  4. cif_plots.py - Generates the cif plot widget. Note that jsmol is used to create the widget