Script for predicting redox errors
It is recommended to use virtual environment so that the required python packages are the same. How to set up python virtaul environment
Afther these are ready change directory to the project and install required Python packages with pip install -r requirements.txt.
The repository containts the main script get_redox_errors.py which is called by the user
tools folder contains helper scripts to parse the given data, adds additional features, and logic to run the inference.
models folder contains all of the models that the script can use. In order to add new models, create a new folder with descriptive name and paste a pickle serialized model inside the folder. For sensors specific models add all 5 pickled models inside the folder. NOTE sensor specific models need to have the word sensor_<number> in the pickle file name to work correctly.
If sensors specific models are used the script gets predictions for each sensor model and combines all sensor findigs as one redox_error_flag by using logical OR on each result.
Minimum way to run the script python3 get_redox_errors.py -f "/path/to/file.csv" -o my_results. All of the script options are defined below.
To get help, run the following command python3 get_redox_errors.py --help. This will output description of the script and guide how to use it.
-f (--file) (Required option). This option defines the path to the file that is fed to the model. Given file type must be csv.
-o (--output) (Required option). This options defines the name of the output file that contains the redox error findings.
-m (--model) (Optional). This option defines the model used for the inference. By default it uses sensor specific SVM models.
-s (--scale) (Optional). This option defines if the data is scaled before the inference. The default value is True