Nonlinear Function Fit

Description

A simple app built with streamlit.io that performs nonlinear function fitting using different methods.

Currently supported methods:

1. Locally weighted regression
2. k-Nearest Neighbours
3. 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.

How to use

Run the app

To run the app, open a command line and use the following command

streamlit run app.py

Features

1. Specify a function, which you would like to approximate
2. Select a specific method to fit the function on some randomly generated training data
3. Predicts the function within a specified range of x-values (via sidebar configurations)
4. Show the predictions of the neural network at different epochs during the training process