interval Holt’s exponential smoothing method

Interval Holt’s exponential smoothing is a method which implements Holt-winter algorithm with interval-valued inputs and interval-valued outputs. This package is based on the paper, Holt's exponential smoothing and neural network models for forecasting interval-valued time series, FAT de Carvalho

Usage

import numpy as np
from iholt import Holt_model
from scipy.optimize import minimize

# Construct the synthetic data
s = np.sin(np.arange(0, 18, 0.1))
s = np.vstack((s, s+0.5))

# split series
s_test = s[:, -10:]
s = s[:, :-10]

# Build model
holt_model = Holt_model(s)

# Optimize
bnds = [[0, 1]] * 8
x0 = np.ones(8) * 0.5   # Parameters [a11, a12, a21, a22, b11, b12, b21, b22]

result = minimize(holt_model.fun, x0, method='L-BFGS-B', bounds=bnds)

print(result)

# predict
It, Lt, Tt = holt_model.pred(result.x, 10, s_test)

kaishxu/iholt

interval Holt’s exponential smoothing method

Usage