Dataset and code used in "Predicting the Superconducting Critical Temperature in Transition Metal Carbides and Nitrides using Machine Learning".
The dataset is contained in the file: dataset.csv. It contains the following columns:
- num: whether the compound was extracted (EXT) or was found in the SuperCon database (the corresponding compound number)
- compound: the compound
- Tc: the experimentally measured critical temperature
- a: the lattice parameter (estimated in some cases, see paper)
- elements: the elements present, therefore the substitution series. (=BASE for pure compounds)
- 1, 2: pure compounds 1 and 2 when the compound is an alloy.
- perc_1_in_2: the fraction of compound 1 in the compound
The dataset_features.csv contains the above-mentioned columns as well as additional MagPie features generated using matminer.
To use the code, Python 3 should be installed on the local machine. A recommended way of installation is through Anaconda, see https://docs.anaconda.com/anaconda/install/index.html.
First clone this repository into the desired directory and access it using the following two commands from a terminal window:
git clone https://github.com/hmetni/Superconducting_TM_C_N.git
cd Superconducting_TM_C_N
Once Anaconda is installed and the repository cloned, create a virtual environment using:
conda create --name Superconducting_TM_C_N python=3.7
Then activate your newly created environment:
conda activate Superconducting_TM_C_N
The next step is to install the requirements. The main packages required for this work are: pandas, scikit-learn, umap, seaborn, matplotlib and matminer.
Once the requirements are installed, the notebooks can be executed in interfaces such as Jupyter Notebook.