Perovskite-Device-Doctor

Identification of the Dominant Recombination Process for Perovskite Solar Cells Based on Machine Learning.

Installation:

To run the Jupyter Notebook, please install the following: Anaconda (https://docs.anaconda.com/anaconda/install/), which already consists of Jupyter Notebook, NumPy, pandas, matplotlib, seaborn, pickle, scikit-learn, joblib.
Jupyter Notebook (https://jupyter.org/install.html)
There are several packages and/or libraries that need to be installed to run the notebook:

NumPy (https://docs.scipy.org/doc/numpy/user/install.html)
Pandas (https://pandas.pydata.org/pandas-docs/stable/getting_started/install.html)
Matplotlib (https://matplotlib.org/users/installing.html)
Seaborn (https://seaborn.pydata.org/installing.html)
Pickle (https://docs.python.org/3/library/pickle.html)
Scikit-learn (https://scikit-learn.org/stable/install.html)
Joblib (https://joblib.readthedocs.io/en/latest/installing.html)
Graphviz (https://pypi.org/project/graphviz/)

Data

Note that the dataset used for the training and testing can be download from Mendeley Data (http://dx.doi.org/10.17632/xbzw29tjz4.1) and consist of over 10⁶ simulated perovskite solar cells with their performance (V_OC,J_SC and FF) at (0.1,0.18,032,0.56,1) sun illuminations as well as the ideality factor (n).
This dataset was built using drift-diffusion open-source code SIMsalabim.
(https://github.com/kostergroup/SIMsalabim)

Running

After installing all the necessary packages run the Jupyter Notebook by running all cells.

Available ML algorithms

The default Jupyter Notebook contains the results for Decision Tree classifier (scikit-learn) and Random Forest Classifier (scikit-learn). However, the code is built such as other types of classifiers can also be used easily.

Author

Vincent M. Le Corre