Time Series Smoothing: an Interactive Visualizer

This repository contains the code for a Streamlit app that visualizes time series smoothing techniques. It includes both real and synthetic datasets and lets you compare how different methods behave with adjustable parameters.

Note: The app was developed while writing this Medium article: Six Approaches to Time Series Smoothing

Features

• Adjustable smoothing parameters
• Visual comparison across methods
• 5 datasets

Supported methods: Moving Average, Exponential Moving Average, Savitzky-Golay, LOESS, Gaussian Filter, Kalman Filter

Datasets

This project uses a mix of real-world and synthetic datasets. Below are the sources and licensing information:

• Sunspots
  Daily total sunspot numbers from SILSO. Licensed under CC BY-NC 4.0.

• Humidity (RH) and Wind Speed (WV)
  Weather time series from Weather Long-term Time Series Forecasting on Kaggle. Licensed under the MIT License.

• Noisy Sine
  Synthetic noisy sine wave, created for this project.

• Process Anomalies
  Synthetic dataset simulating different industrial operating modes and injected anomalies, created for this project.