Quick Help for "Web Decomp"

(This page is in preparation now!)

In this system, you can analyze your time series data by using popular methods such as AR model, ARMA model, Seasonal Adjustment Method (by Decomp) and etc. All computation will be done in our server machine. You do not need any installations or registeration.

The main method of this system is "Decomp" which was proposed by Kitagawa and Gersch (1984). Decomp is the seasonal adjustment method using State-Space-Modeling. You can obtain not only the adjusted series but also the very smooth trend series.

Any suggestions, comments or questions are welcome! Please contact me(sato@ism.ac.jp).

1. Input your data

• Currently, only univariate data can be analyzed. You can input your data in the input field of the first page from the keyboard. But it is recommended to use "Cut & Paste" by your mouse when the number of the data is large. You should input only the values. Do not contain any character in this field. These data must be separated by "space", "comma" or "return".
• If you need, you can input the title of the data in "title" field. But this is not necessary. (You can input only alphabets and numbers except ".".)

2. Run

• Select the method that you want to do from the menu "Select Method". --- List of the Methods
• If the selected method has some options, set the parameters below. (Some methods do not have any parameter.)
• Press "Run".

3. Result

• When the computation is completed, the result will be displayed in your browser automatically.
• There are some graphs and the numerical results. If there are some data outputs, these are input in the data fields below the parent window well as original data.
• If you want to continue another analysis using output data, press the button of the output data which you want to use. Then these data are copied to "data input field". You can analyze the output data, or apply another method to the original data quickly.

Reference

Kitagawa, G., and Gersch, W. (1984), "A smoothness priors - state space modeling of time series with trend and seasonality", JASA, Vol.79, No.386, 378-389.

(This page will be revised as soon as possible. This is not complete yet.)

Go to Main Page

Sato's Home Page

Go to Guest Book

sato@ism.ac.jp