Data processing device, data processing method, and program
A data processing device includes: a data obtaining section obtaining time series data on a total value of current consumed by a plurality of electric apparatuses; and a parameter estimating section obtaining a model parameter when states of operation of the plurality of electric apparatuses are modeled by a factorial HMM on a basis of the obtained time series data.
1. A data processing apparatus comprising:
a data obtaining section configured to obtain time series data on a total value of a plurality of waveforms or vector time series data;
a parameter estimating section configured to obtain a model parameter when states of said plurality of waveforms or vector time series data are modeled by a stochastic model on a basis of said obtained time series data or said vector time series data, wherein in a parameter estimating process based on an algorithm which alternately calculates expectation and maximizes likelihood, under constraint conditions that converted values or a vector of an observation parameter W(m) corresponding to a pattern of a waveform of a factor m of said stochastic model are nonnegative, said stochastic model maximizes a likelihood function as a degree of said stochastic model describing a pattern of the total value of said waveforms indicated by said time series data, whereby said parameter estimating section obtains the observation parameter W(m) as said model parameter;
a state predicting section predicting a state of said factor at a future time after passage of a predetermined time from a present time; and
an apparatus control section controlling a predetermined apparatus based on a result predicted by state predicting section.
2. The data processing apparatus according to claim 1 , wherein, in said parameter estimating process based on said algorithm, under constraint conditions that a number of states assumable by each state of said stochastic model is two, said stochastic model maximizes said likelihood function as a degree of said stochastic model describing said pattern of the total value of said waveforms indicated by said time series data.
3. The data processing apparatus according to claim 1 , wherein, in said parameter estimating process based on said algorithm, under constraint conditions that one of a first column and a second column corresponding to said number of states is zero at all times, said stochastic model maximizes said likelihood function as a degree of said stochastic model describing said pattern of the total value of said waveforms indicated by said time series data.
4. The data processing apparatus according to claim 1 , further comprising an identical determining section determining that a plurality of predetermined factors of said stochastic model correspond to the time series data.
5. The data processing apparatus according to claim 4 , wherein said identical determining section selects combinations that satisfy a necessary condition at all times among all combinations of each factor of said stochastic model as identical candidates, and determines that a combination that has satisfied a hint condition a predetermined number of times among the selected identical candidates corresponds to the time series data, whereby said identical determining section determines that the plurality of predetermined factors correspond to the time series data.
6. The data processing apparatus according to claim 1 , further comprising an apparatus identifying section identifying a plurality of apparatuses corresponding to each respective waveform.
7. The data processing apparatus according to claim 6 , further comprising a display unit displaying a state of said plurality of apparatuses identified by apparatus identifying section based on the result estimated by said parameter estimating section.
8. The data processing apparatus according to claim 1 , wherein said data obtaining section obtain time series data on a total value of waveform on predetermined cycles synchronizing phase of a measured value.