From MAGGIE

[edit] Introduction

The purpose of this Project is to facilitate the network managers by providing them with the predictions of network behavior in future. Such forecasts can ensure the efficient utilization of network resources. In this project, we are required to analyze the time series data of various network parameters on which respective forecasts are to be based. Modeling of forecasting the time series data is referred as "Time series Analysis & Modeling" in the literature. The general steps involved in Time series forecasting are as follows:

1. Obtain the data
2. Preprocessing (Preliminary Transformations)
3. Eliminate the Deterministic Components
   1. Trend
   2. Seasonal Component
4. Choose a model to fit the Random Component
5. Forecast the next Data

[edit] Elimination of Trend

Studies reveal that if the time series data is not fitted well to a polynomial of degree then there exists seasonality in time series data. In estimating the time series data to a polynomial, the sum of squares of the residuals(also called sum of squares of errors) is used in the estimation equation. <math>residual = Observed Data - Estimated Data</math> We assume that the data can be fitted to a polynomial of degree three. Image:D:\Research\Time Series Analysis\Presentations\Editing\estpoly.gif