Advances and Applications in Statistics
Volume 53, Issue 6, Pages 731 - 770
(December 2018) http://dx.doi.org/10.17654/AS053060731 |
|
POTATO PRICE FORECASTS IN DELHI RETAIL FROM BAYESIAN REGIME SWITCHING
Olli Salmensuu
|
Abstract: A Bayesian regime switching model for month-ahead forecasts of potato price in Delhi retail trade is constructed. The monthly crashing limit is set to -10 percent. The price crashes annually due to main harvest, starting in either November or December. Using same limit, either January or February is still crashing in the harvest aftermath. The regime switching both at the beginning and end of the crash is thus a Bernoulli trial that here updates conjugate Beta distributions. Same structure for Bayesian learning is set also for crash oscillations between lower and upper bottoms with Markov Chain dependency on the previous bottom type updated probability. Rising of the price in between the crashes is implemented by conjugate normal distribution and previous month data. |
Keywords and phrases: potato price, retail price, Bayesian model, conjugate prior, regime switching.
|
|
Number of Downloads: 379 | Number of Views: 3261 |
|