Logistic regression model on land productivity of Pagar Alam coffee farming

Abstract

Pagar Alam coffee farming is a smallholder plantation, the majority of which is a hereditary business. The success of this coffee farming cannot be separated from existing resources, including land productivity. Land productivity concerns the amount of production, land resources, and land management efforts. This paper discusses the factors that influence the land productivity of coffee farms in Kota Pagar Alam, using binary logistic regression analysis. In general, there are 5 factors discussed, namely the identity of farmers and their internal factors, agricultural land, the performance of farmers in the production process, yields, and external factors on the productivity of Pagar Alam coffee farms. The data used are 191 respondents with 33 independent variables and one dependent variable. Each variable is divided into categories. Land Productivity as the dependent variable is divided into 2 categories, namely low and high. Based on bivariate analysis, variables related to land productivity are land area, number of trees, frequency of fertilizer used, frequency of pesticides used, length of harvest, production, female labor in the family, gross income, net income, and production costs. Furthermore, based on the binary logistic regression model of land productivity probability, variables that significantly affect land productivity of Pagar Alam coffee farms are area, number of trees, crop production, and net income. The accuracy of the model simultaneously was 93.2%. The probability value of the model is predominantly influenced by the harvest production variable with an odds ratio of 49.505. If the category of harvest production and net income increases, the probability for high land productivity will also increase. Conversely, if the area of land and the number of trees increases, the probability of high land productivity will decrease.