A tool to make virtual sensors for data quality assessment
- Python 3.8
- matplotlib
- numpy
- pandas
- plotly
- scikit-learn
- Tensorflow 2.0
- pyyaml
- scipy
- dvc
Stages:
- Select: Selects usefule features of raw data into a new dataframes.
- Transform: Transforms dataset (at the moment add zeros). . Split: Split data set into training and test data.
- Scale: Scale input data.
- Sequentialize: Split data into input/output sequences.
- Train: Train model.
- Evaluate: Evaluate model.
All stages are defined in the file dvc.yaml
, and the parameters to be used
are saved in params.yaml
.
To run/reproduce an experiment with any given parameters specified in
params.yaml
, run:
dvc repro
To run experiments with another dataset, just change the content of
Depot/data/raw/CNC_Milling_dataset
to the files you want to use.