Master's Thesis Source Code - Automated Data Analytics for Motifs and Discords Mining
These instructions will get you a copy of the project up and running on your local machine.
See requirements.txt for the required Python packages.
Make sure have the HDF5 data of the REDD or UK-DALE dataset into a data directory. For example, data/REDD/redd.h5 or data/UKDALE/ukdale.h5.
You also can run the code through the command line interface without having to start the Flask server.
You must specify the following arguments:
--datapath Data file path
--train_building Train building
--train_start Train start date (Optional); Default = start datetime of dataset
--train_end Train end date
--val_building Validation building
--val_start Validation start date
--val_end Validation end date
--test_building Test building
--test_start Test start date
--test_end Test end date (Optional); Default = end datetime of dataset
--appliance Target appliance
--sampling_rate Downsampling rate
--epochs Max number of epochs for Neural network approaches
--patience Number of iters Neural networks allowed to continue training with increase in val_loss
--max_evals Number of max trail to run
--metrics_to_optimize Metric to optimize for
Example of execution of the script with the arguments
# change into directory with the cli script
cd bayesian_optimization
# run the script with specified settings
python automl_hyperopt_cli.py --datapath ../data/REDD/redd.h5 --train_building 1 --train_end 2011-05-09 --val_building 1 --val_start 2011-05-10 --val_end 2011-05-17 --test_building 1 --test_start 2011-05-18 --appliance fridge --sampling_rate 120 --epochs 10 --patience 5 --metrics_to_optimize "mean_squared_error" --max_evals 10
The result will be kept in bayesian_optimization/results/
For each run of the AutoNIALM, the code with output two json files:
- One for the metadata of the results (
trial-metadata-[APPLIANCE]-[DOWNSAMPLING RATE]-[METRIC TO OPTIMIZE].json) - Another file containing information on every trial (
trial-results-[APPLIANCE]-[DOWNSAMPLING RATE]-[METRIC TO OPTIMIZE].json).
Note: Currently running the GUI on flask server doesn't support using validation sets. This will be updated soon.
# make the script executable
chmod +x flask_run.sh
# run the script to start flask server
./flask_run.sh
This project is licensed under the Apache License 2.0 - see the LICENSE.md file for details
- Neural Network approaches adapted from: https://github.com/OdysseasKr/neural-disaggregator