Paper:"Comparison and development of machine learning tools for the prediction of chronic obstructive pulmonary disease in the Chinese population"
1: data_clean.ipynb
Split <the 20191023_finalalldata.csv > to train data and test data (train_data.csv & test_data.csv)
2: copd_trainset_final_code.ipynb
Use the machine learning method(SVM，LogisticRegression，DecisionTree，MLP，Xgboost，KNN) with CV =5 to select the model and get average value of different evaluation parameter.The results are 5-fold tables, AUROC and PRROC with 5-fold labels, respectively.
3: copd_validate_final_code.ipynb
Validate the 6 models(from the train data) in test_data and set the model parameter to get the best result. The results are tables, AUROC and PRROC,respectively.
4: data_1024 (input file)
the 20191023_finalalldata.csv
5: result_1024 (result file)
feature_importance (plot):only use the xgboost method to get the all feature importance.
allfeature file: Result from above (2,3 code) include the 5 clinical features and 9 SNPs OR values.
clinical file: Result from above (2,3 code） just only include the 5 clinical feaures.
feature_snp file: Result from above (2,3 code) just only incule the 9 SNPs OR values.