統計学輪講（第25回）

日時      2012年01月24日(火)    15時00分～15時50分
場所      経済学部新棟３階第３教室
講演者    大山 遼 (経済M2)
演題      Bayesian analysis of TEPCO electricity load data using state
space model

概要
Finding an appropriate model that describes the structure of the time
series of the electricity load data is of great importance. From
statistical point of view, less attention has been given to the application
of the latest methods to this field compared to other fields such as
finance and macroeconomics.
Some previous studies have discussed the relatively simple linear Gaussian
state space models, but extending this to the more flexible non-Gaussian
state space model can be readily done by mixing Gaussian distributions. We
will analyse the time series of the maximum electricity load within the
jurisdiction of the Tokyo Electric Power Company (TEPCO), introducing
several components such as the meteorological effect terms and the day type
terms.