IMPLEMENTATION OF TIME SERIES FORECASTING USING GENERALIZED ADDITIVE MODEL IN WEB APPLICATION

Abstract

Analyzing a huge amount of data is a mundane task as it needs expertise in the area, a lot of concentration for repetitive tasks especially in time series data. Using Statistical Modeling that is general enough for any types of time series data could improve the flexibility of anyone who wants to analyze the time series data, as long as they have the domain knowledge on the time series data. The implementation of web application could improve the accessibility for anyone who wants to analyze time series data. The Statistical Modeling were using the method of Generalized Additive Modeling for the Decomposed Time Series because of its flexibility in forecasting the trend, seasonality, and other regressors for different time series which can help anyone with the domain knowledge for analyzing time series data. After some iteration of observing 5 months data of the Stock Price movement and predicting the next week‘s Bullish and Bearish trend for every week, the result shown 68% accuracy for every Bullish and Bearish trend. While in calculating the forecasting errors, the result shows a reliable prediction with no more than 0.015 in terms of Mean Absolute Error value for predicting the next seven days for every sixty days.