Modeling effects of climatic variables on tea production in Kenya using linear regression model with serially correlated errors

Abstract

Aims/ Objectives: To formulated a linear regression model to capture the relationship between tea production and climatic variables in terms of ARIMA. Place and Duration of Study: Department of Mathematics and Actuarial Science, Catholic University of Eastern Africa, Nairobi, Kenya, between June 2019 and April 2021. Methodology: The study used time-series data for mean annual temperature, mean annual rainfall, humidity, solar radiation, and NDVI, collected from six counties, namely Embu, Kakamega, Kisii, Kericho, Meru, and Nyeri. Results: The study findings noted that there is a presence of trend and seasonality for all the data. The scatter plot matrix for all the climatic variables for all the counties under the study indicated that tea production has a linear relationship with most climatic variables. Model fit of the data indicated statistical significance when tea production data is differenced. A second linear model with tea production data deseasoned has mixed results in terms of a significance test. The variation of independent variables with tea production yielded very low values, suggesting that the data used has many variabilities. Conclusion: The study findings show the climatic variables can be used to forecast tea production. Recommendation: Future studies may combine the analysis with other statistical modeling procedures such as the GARCH models.