Example on sparse_lagged implementation
Sparse lagged implementation was overwritten with flag of using svd or not. In case of simple sparse, features matrix thinns out with the percent of components to save as a hyperparameter evenly. Simple way was added because of increasing time usage on short time series.