An Approach Of Pls Method Applied To Model The Rice Self-Sufficiency Of Peasant Households In Atsinanana Madagascar

The aim of our study is to build a predictive model using data from survey studies. As the predictors were mostly qualitative variables, we applied an extension of PLS method. This work responds also to the need expressed by the Atsinanana agricultural region of Madagascar to strengthen the decision-making process in the fight against "no-self-sufficiency in rice" and the poverty in this locality.The study consists of constructing a simple scientific tool to improve the comprehension and to explain the links between permanent poverty and the main socio-economic factors: schooling, illiteracy, fragmentation of land, insecurity, isolation, demography) and to make effective the measures taken.The data is a random sample from a database collected in the 82 rural communes of Atsinanana by the students of the University of Toamasina and from the Campus Paysan project. We propose a Partial Least Square regression model of rice self-sufficiency.This study has helped to update the information on the subject and to deepen the knowledge on the issue. It is part of a search for improvement of statistical prediction methods by the PLS regression. We worked mainly with free software R.