This Machine Learning model is helpful in aiding home automation and home appliaances when there is limited supply of power. Typical home appliances are considered and power consumption of these devices are noted down based on properties like type of device, time (hourly basis), power consumption, number of people and time they have used the appliances in an hour.
- Python 3.x
- Numpy, Scipy, Pandas, Matplotlib
- TensorFlow 1.x+, Keras 2.x+
- NVIDIA CUDA 9.x, cuDNN 7.x (Only if using Tensorflow-GPU)
- Sci-kit learn
home
│ README.md
│
└───data
│ sample - contains sample dataset
│ trial_x - all types of different datasets
│
└───regression
│ *.ipynb - Regression ML/DL algorithms
│
└───model
│ regression models saved
│
└───docs
│ documentation and reports
│
└───main
visualization - data analysis
data generation - generate data and cleaning
experiment - results and custom algorithms
-
main/data generation/generate_home_data.py
- generateshome_data.csv
- Generates dataset with all columns
- This dataset will be used to build machine learning model
-
main/data generation/generate_home_data_test.py
- generateshome_data_test.csv
- Generates dataset with all columns
- This dataset will be used to predict power values
-
main/visualization/total_power_consumption.ipynb
- createsdate_time_group.csv
- This is the final output which will be used to visualise and compare machine learning algorithms
-
main/visualization/total_power_consumption_final.ipynb
- createsfinal_output_group.csv
- This is the final output which will be used to visualise and compare machine learning algorithms
-
regression/<*>.ipynb
- Machine learning algorithms used on
home_data.csv
to build models in/model/
- Machine learning algorithms used on
-
predict/<*>.ipynb
- Machine learning algorithms used on
home_data_test.csv
to predict and generate new datasets home_data_predict_x
are dataset generated with prediction values
- Machine learning algorithms used on
-
main/experiment/trail_x/trail_x_x.py
- Custom algorithm to give messages to users - generates
final_output_x.csv
indata/trail_x
- Custom algorithm to give messages to users - generates
-
main/experiment/trail_x/results_x_x.py
- Visualising all algorithms and performances
- Graphs plotted to identify power consumption before machine learning and after using machine learning and custom algorithm
Available in main/experiment/trail_x/results_x_x
where 'x' can be either sequence or a machine learning algorithm
CTO, Faststream Technologies
Copyright(c) 2018, Faststream Technologies
Authors: