table15

Table1.5

Table 1.5 is a Python application that can generate a table that is adjunct to a typical Table 1 (association statistics). Table 1.5 goes beyond static association by analyzing the impact that a change in each single feature has to changes in the outcome.

Installation

Use the package manager pip to install Table1.5.

pip install table15

Code Structure

table15
├── LICENSE
├── README.md
│       ├── __init__.py
│       ├── magec_sensitivity.py
│       ├── mimic_queries.py
│       ├── mimic_utils.py
│       ├── rbo.py
│       └── table15
│           ├── __init__.py
│           ├── __main__.py
│           ├── magec_utils.py
│           ├── pima_utils.py
│           ├── pipeline_utils.py
│           ├── runner.py
│           ├── utils
│           │   ├── __init__.py
│           │   ├── data_utils.py
│           │   ├── magec_utils.py
│           │   ├── model_utils.py
│           │   ├── pima_utils.py
│           │   └── pipeline_utils.py
│           └── viewer.py
├── pyproject.toml
├── src
│   ├── __init__.py
│   ├── data
│   │   ├── diabetes.csv
│   │   ├── healthcare-dataset-stroke-data.csv
│   │   └── linear_data.csv
│   ├── table15
│   │   ├── __init__.py
│   │   ├── __main__.py
│   │   ├── configs
│   │   │   ├── data_configs
│   │   │   │   ├── pima_full.yaml
│   │   │   │   ├── pima_lite.yaml
│   │   │   │   └── stroke_full.yaml
│   │   │   ├── model_configs
│   │   │   │   ├── deep_models_configs
│   │   │   │   │   └── multi_layer_perceptron_1.yaml
│   │   │   │   ├── ensemble_configs
│   │   │   │   │   ├── random_forrest_cc_1.yaml
│   │   │   │   │   └── voting_classifier_1.yaml
│   │   │   │   ├── linear_model_configs
│   │   │   │   │   ├── lr_1.yaml
│   │   │   │   │   ├── lr_2.yaml
│   │   │   │   │   └── lr_cv_1.yaml
│   │   │   │   └── svm_configs
│   │   │   │       ├── linear_svm_cc_1.yaml
│   │   │   │       └── svm_1.yaml
│   │   │   └── pipeline_configs
│   │   │       ├── linear.yaml
│   │   │       ├── pima.yaml
│   │   │       ├── stroke.yaml
│   │   │       └── synth_data.yaml
│   │   ├── configs.py
│   │   ├── models
│   │   │   ├── __init__.py
│   │   │   ├── deep_models.py
│   │   │   ├── ensemble_models.py
│   │   │   ├── linear_models.py
│   │   │   ├── model.py
│   │   │   ├── model_factory.py
│   │   │   ├── svm_models.py
│   │   │   └── test_linear_model.py
│   │   ├── perturbations
│   │   │   ├── __init__.py
│   │   │   ├── group_perturbation.py
│   │   │   ├── perturbation.py
│   │   │   └── z_perturbation.py
│   │   ├── runner.py
│   │   └── utils
│   │       ├── __init__.py
│   │       ├── data_tables.py
│   │       ├── magec_utils.py
│   │       ├── models_container.py
│   │       └── pipeline_utils.py
│   └── table15.egg-info
└── tests
    ├── configs
    │   └── t_configs.yaml
    ├── test_perturbations.py
    └── test_pipeline_utils.py

Usage

To generate Table 1.5, the main entry point is through runner.py. This takes a single parameter pipeline_configs_path, which contains arguments to run the pipeline, as well as references to other configs (Data Configs, Model Configs) that are necessary to run the pipeline. Generate a Pipeline Configs yaml (example at src/table15/configs/pipeline_configs/stroke.yaml) that contains general parameters for the pipeline that are meant to be changed frequently for different runs. Reference a Data Configs yaml (example at src/table15/configs/pipeline_configs/stroke.yaml) that references configs related to data and are meant to be more static (ie, we don't change the data arguments very often) Also reference a set of Model Configs yamls (example at src/table15/configs/model_configs/linear_model_configs/lr_1.yaml), which almost never change except for tuning and when implementing a new model. Together, these can be used to generate Table 1.5

import table15
table15.runner.run(`path_to_pipeline_configs_yaml`)

Support

Issues and support can be directed to @KaleRP

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

Authors and acknowledgment

Author of this project is @KaleRP. Special thanks to @gstef80 for creating the original project this application was forked from. Another special thanks to @beaunorgeot for originally conceiving this project.

License

MIT