A simple app built with streamlit.io that performs nonlinear function fitting using different methods.
Currently supported methods:
- Locally weighted regression
- k-Nearest Neighbours
- Neural network
Each method is being trained on a randomly generated dataset (with added noise) and evaluated on equally distanced samples of a specified range of x-values. The respective results are plotted in a plotly-generated graph.
To run the app, open a command line and use the following command
streamlit run app.py
- Specify a function, which you would like to approximate
- Select a specific method to fit the function on some randomly generated training data
- Predicts the function within a specified range of x-values (via sidebar configurations)
- Show the predictions of the neural network at different epochs during the training process