This is an unofficial implementation of “A Stacking-Based Model for Non-Invasive Detection of Coronary Heart Disease” paper. This project consists of all parts of experimental and proposed methods and you can compare their results with reported results. According to evaluations there is no abnormality in reported and evaluated results. (This implementation is a part of research project and tries to present this king of solutions and find out some new approaches to handle these kinds of problem states.)
this project has four main parts consist of Data Pre-processing, Feature selection, Proposed methodology and Experimental results. Functions are created separately for all of mentioned parts and you can use them according to ‘all_functions_run.ipynb’ notebook guidance. This project tries to induct you with different methodologies in simple way and open your mind to make new opportunities. If you have any problem to understanding codes, please contact us through this email address:
mehrzadiarash@gmail.com
all results were recording in separated folders in ‘results’ according to your system date and time so you can check the results after each running. Different results stored in separated .xlsx files so you can find them by the name of your calling function. Following chart is create to find out the naming method.
| Section | Title |
|---|---|
| table 2 | datapath_processed.xlsx ' and ' datapath_rowdata.xlsx |
| table 3 & 4 | datapath_processed.xlsx CHI2_fs_dataset.xlsx RFECV_fs_dataset.xlsx XGB_fs_dataset.xlsx mutual_fs_dataset.xlsx variance_fs_dataset.xlsx SVC_fs_dataset.xlsx feature_selection_RES.xlsx featureselection_RES_Kvalue.xlsx |
| table 5 | bigdatapath_processed.xlsx |
| table 6 | report_paper_model_RFECV.xlsx |
| table 9 | report_paper_model_bigdata.xlsx |
All functions are implemented without any complexity and extra parameters. All required parameters was define in config files and you can easily call your expected function as same as presented ways in ‘run.ipynb’ files. All parameters were set as same as reported parameters in mentioned paper and for some unknown values parameters, default values were set. You can easily change those parameters in relevant config files and trying different ways to learn models.
Sadly, an official source code is unavailable and we can’t evaluate our implementation with official method but according to evaluations, our implementation gains a close result to official results that was reported in mentioned paper. You can get mentioned paper in ‘‘IEEE Access’’ journal and its topic is “A Stacking-Based Model for Non-Invasive Detection of Coronary Heart Disease”. Link → ('paper’). To access the datasets related to this research, you can check following chart and visit mentioned links.
| Dataset | Link |
|---|---|
| Z-Alizadeh Sani Data Set | Link |
| Cardiovascular Disease dataset (big data) | Link |