My goal was to create a project that had some meaning. This to me is meaningful as it predicts heavy machinery failure types. In the world of mining and heavy industry, predictive maintenance techniques are golden as they can save a business a lot of time and money.
The dataset used here is synthetic but fairly accurate to groomed data you'd see collected from mobile assets. It is however missing a lot of contextual data that makes classifying and determining failures more meaningful and concrete.
If you want to skip to the end result of my experiments, have a look at this notebook: Final_Project_ML_Neural_Net_v2.ipynb.
- Data analysis
- PCA
- Statistical analysis
- Decision trees
- Neural networks
- Colab
- Sklearn
- Pandas
- Matplotlib
- TensorFlow
- Keras
- OpenML