Coffee Price Determinants

Project Overview

• Dataset: 1994–2022, 29 years, 13 variables
• Goal: Identify key drivers of global coffee prices
• Tools: R, tidyverse, lm(), stepwise AIC, ANOVA, relaimpo

Key Findings

• Top drivers: Imports (−), Exchange rate (−), World Population (−), Domestic Consumption (+, borderline)
• Adding Year improves fit (ANOVA p = 0.022), capturing time trends
• With Year: R² = 0.87; Without Year: R² = 0.83 → stronger residual stability with Year

Model Diagnostics and Visuals

Sample Code (R)

# Compare models
model_with_year <- lm(`Coffee Price` ~ Year + `Exchange rate` + `World Population` + `Domestic Consumption` + Imports, data = train)
model_without_year <- lm(`Coffee Price` ~ `Exchange rate` + `World Population` + `Domestic Consumption` + Imports, data = train)

# ANOVA comparison
anova(model_without_year, model_with_year)

# Variable importance
library(relaimpo)
calc_relimp(model_with_year, type = "lmg", rela = TRUE)
  

Contribution to the International Coffee Organization (ICO)

I contributed to the ICO Global Coffee Funding Mechanisms report (June 2024), by writing the boxes and providing insights on market dynamics and price drivers in the coffee industry.