What is l1 trend filtering?
Like H-P filtering, ℓ1 trend filtering is a batch method for estimating the trend component from the whole history of observations.) We will see that the proposed ℓ1 trend filter method shares many properties with the H-P filter, and has the same (linear) computational complexity.
What is trend filter?
The goal of trend filtering is to smooth out a time series by filtering out the 'noise'. The trend filtering algorithm faces a trade-off between two objectives. Firstly, it wants to minimise the residual 'noise' between the actual and smooth series.