This is a project about predicting the influence of companies, using the performance on GitHub: h index.
If the amount of api calls is limited, you can use your own token. (The token creation method is written in common.) If the code runs breaks, you can use a cache.
no2_hist.py plots the company's social network, together with bars for each centrality metric. (Use common contributors as a basis for connecting edges.)
no1_plot_roc.py is a prediction model using 11 features and XGBoost, and draw the roc curve.
no3_svm_f1.py is a prediction model using 11 features and svm (takes a long time to run, please be patient)
no4_mlp_f1.py is a prediction model using 11 features and mlp.
Crawled data for all companies is stored in json format in data_jsons.zip
All the output images are stored in output.
This project crawled 486 companies' information, if you only want to use some of the companies data, you can modify in data.