Improvisation on Forecasting Algorithm for Time Series Data with Removal of Outliers

A time series is a sequence of observations y1… yn. We usually think of the subscripts as representing evenly spaced time intervals (seconds, minutes, months, seasons, years, etc.). Time series analysis is a statistical technique that deals with time-series data, or trend analysis. Time series data means that data is in a series of time periods or intervals. Several algorithms are available to forecast from time-series data, such as Moving Average, Exponential smoothning, Double Exponential, Triple Exponential, ARMA, ARIMA and some of the other customized algorithms as well. Existing algorithms consider exact value of an attribute of input data for calculation. No such algorithm has the functionality which can identify outliers from input data (if exist) and can convert the outlier into the non-outlier with calculation (increment/ decrement) before performing the forecasting steps. My study concentrates on getting a solution/an algorithm which can identify outlier (if any) as per the parameter(s) (applied at initial level) and filter out these outliers from input data or can convert them into non-outlier value.