Agricultural Research Management Information System

Home
Research Summary
Search
About Us
- ARMIS
- Brochure
Contact Us
Report
User Request
Data Input
Help
- Operation Manual
  - PDF
  - Video
- Program Area & Commodity
We have reached 37600 number of research entries at this moment.

- Logout

Research Detail

Home
Research
Detail

Market Model Analysis and Forecasting Behavior of Watermelon Production in Bangladesh

Murshida Khanam
Department of Statistics, Biostatistics & Informatics, University of Dhaka,Dhaka-1000, Bangladesh

Umme Hafsa
Department of Statistics, Biostatistics & Informatics, University of Dhaka,Dhaka-1000, Bangladesh

Abstract/ Executive Summary:

An attempt has been made to study various models regarding watermelon production in Bangladesh and to identify the best model that may be used for forecasting purposes. Here, supply, log linear, ARIMA, MARMA models have been used to do a statistical analysis and forecasting behavior of production of watermelon in Bangladesh by using time series data covering whole Bangladesh. It has been found that, between the supply and log linear models; log linear is the best model. Comparing ARIMA and MARMA models it has been concluded that ARIMA model is the best for forecasting purposes.

Keywords: ARIMA, MARMA, Supply model, Log linear model

Location: Department of Statistics, Biostatistics & Informatics, University of Dhaka

Start Date: 00-00-2012

End Date: 00-00-2012

Program Area: Socio-economic and Policy

Commodity: Water melon

Objectives:

1. In this study it has been found that ARIMA model can be applied to forecast the demand and price which are highly unstable.

Methodology:

By using the concept of previous literature, the present study has been tried to fit different types of models. This study was conducted to fit the supply and the log linear models and to identify the best one. Regarding forecasting purpose, the present study also considered the ARIMA and MARMA models and to identify the best model that may be used for forecasting purposes. Data: This study has considered the annual time series data for the period 1990-1991 to 2009- 2010 covering whole Bangladesh from various publications of BBS (Bangladesh Bureaus of Statistics) and the price of watermelon has been collected from BADC (Bangladesh Agricultural Development Council), Khamar Bari, Farmgate, Dhaka. Variables of the study: Realizing the limitations of data availability, three variables have been considered. Production of water melon is taken as the dependent variable for all models where price has been considered as independent variable for supply, log linear and ARIMA models. In MARMA model price, distribution of improves seeds and natural calamities have been considered as independent variable.

Source of Information: Bangladesh J. Sci. Res. 26(1&2): 47-56, 2013 (December)

Article Link (Online Journal):

Funding Source:

Total Budget (Tk.) :

Achievement/Findings:

According to the process of estimation of a model we have estimated our parameters of the models and calculated their respective R² and adjusted R² to compare the models. It has been observed that, for supply model R² which is 0.623 and is good enough that means the model is well fitted. On the other hand, log linear model also fitted well because of higher value of R² which is 0.714. For both the cases we have got significant result of parameters. Comparing the results of R² and adjusted R²of supply model and log linear model it can be concluded that log linear model is better than supply model because the value of R²and adjusted R² of log linear model is greater than supply model. To compare the models we have used R² and adjusted R² as they are the measures of goodness of fit.

Knowledge Management: Journal