Bayesian Predicted Goal for Vegetable Waste Reduction

Github Link

This code analyzes vegetable waste by grade and week using Bayesian modeling in Stan, with R for data prep, prior estimation, and visualization. It predicts waste probability, compares observed vs. predicted values, and plots trends over weeks and grades.

Purpose

Update vegetable waste predictions using posterior data.

Statistical Techniques:

• Bayesian inference with Stan (Models uncertainty in waste probability).
• Beta distribution for waste probability (Fits 0-1 ratio data, versatile shapes).
• Logistic regression for Grade and Week (Models decreasing trends with S-curve).
• Prior estimation from mean and variance (Sets grade-specific priors).
• MCMC sampling for posterior predictions (Handles complex distributions, uncertainty).

Output

• Enhanced waste trend predictions by week and grade.
  1. Change Over Weeks (Mean Predicted Values)

1. Separate Graphs for Each Grade

2. Compare Observed vs Predicted Values (Mean by Grade)