METHOD AND PROGRAM FOR DETECTING CHANGE-POINT OF TIME SERIES DATA, AND METHOD AND PROGRAM FOR PREDICTING PROBABILITY DENSITY DISTRIBUTION OF FUTURE TIME-SERIES DATA VALUES
The present invention applies the particle filter method to a PUCK model that is for calculating the real market price P(t+1) at time(t), which is determined by the real market price P(t) and a median price PM(t). First, particles having parameters are generated, said parameters indicating the states of the PUCK model and each of which having different values, to obtain a probability density function of the parameters. Then, likelihood of each of the particles is evaluated, and resampling of the particles is executed according to the likelihood, as follows. A random number is generated and this random number is compared to a prescribed value, and when the random number is greater than the prescribed value, a particle is regenerated according to a probability density function such as a normal distribution that has the parameter value of the model at time (t) as the average value thereof, and when the random number is less than the prescribed value, a particle is regenerated using a uniform distribution as the probability density function. A series of these processes is to be continued.