Implementation of Smoothing Strategies Combined with ARIMA and Neural Networks to Improve the Forecasting of Traffic Accidents
import numpy as np from tfilter import hsvd N = 500 x = np.sin(np.arange(N) * np.pi/50.0) x = x + np.random.normal(0, 0.3, size=N) window = 100 rank = 2 low_freq, high_freq = hsvd(x, window, rank)Replacing SVD with NMF
time series data must be non-negative
import numpy as np from tfilter import hnmf N = 500 x = np.sin(np.arange(N) * np.pi/50.0) x = x + np.random.normal(0, 0.3, size=N) x = x + 2.0 assert(np.min(x) > 0.0) window = 100 rank = 3 low_freq, high_freq = hnmf(x, window, rank)python tfilter.py
