Abstract: The study on the Forecasting of Monthly Prices of Onion in Major Markets of Maharashtra is selected with an objective estimate the better forecast for the monthly onion prices and test validity of model. For this Support Vector Regression is used to classify the data and fitting better non-linear line. With this the length of support vector can be obtained and it can be use to forecaste the price. For the study, around 14 market price data for are taken from Maharashtra. The data is collected in the form of model price per day from 2002 to 2012 from the website of agmarknet and further converted to monthly price.
The prices forecasted by Polynomial and Sigmoid model of Support Vector Non-Linear Regression are very much closer with actual prices than ARIMA forecasted prices. Deviation declined much more between actual value and forecasted value. As the price of onion per quintal in actual market was Rs.298 in Lasalgaon market and forecasted price is Rs.313 and Rs.314. If the methodology by putting volatility and better optimisation can be tried, there are chances of better results. The result shows that if the data is classified by Support Vector Regression, the forecasted prices are better over other models. The validity of forecasting model is estimated by AIC, R Square and DW test.
From the results of the model fitting, it can be concluded from above results that the model fitting is good in all market except two markets. The best fitted model is observed in Mumbai, Solapur, Nagpur and Pune sequentially.
Keywords: ARIMA forecasted prices, Markets of Maharashtra, monthly onion prices, Support Vector Regression.
Title: Forecasting Of Onion Prices in Maharashtra: An Approach to Support Vector Regression and ARIMA Model
Author: Dr. Sangita Vishnu Warade
International Journal of Social Science and Humanities Research
ISSN 2348-3164 (online)
Research Publish Journals
