This is a python library for multi-seasonal time series regression with Fourier features and smoothing.

MSFR (Multi-Seasonal Fourier Regression) was created to address the limitations of traditional polynomial regression in modeling periodic data.
High-order polynomials often lead to overfitting and unstable forecasts, while ignoring the rich seasonal structures present in many real-world time series.

By incorporating sinusoidal functions as features, MSFR captures both low and high-frequency patterns with fewer parameters,
providing smoother and more interpretable forecasts for seasonal and multi-seasonal data.