Top Page | Upper Page | Contents | About This Site | JAPANESE

Condition Analysis

I call "Condition Analysis" that it uses time-series data as the data set like multi-variable analysis. And it is a kind of time series analysis. It does not use the information of time in the data.

It needs the treatment of outlier and missing values. But it does not need other changes of data. So it is easier than other methods of time series analysis.

Worsening of the Model

If we can make a good model, the model will worsen. This problem is similar to Failure of Time Depended Extrapolation.

Step for Worsening

It is too difficult to make a perfect model that can be used permanently. So we brush up the model.

MW-model (Moving Window) and JIT-model (Just-In-Time) define the range of time of the data to make the model. And they use new data.

NEXT Time Analysis