Capstone internship at the Samsung Research Philippines with a mini-study on : CAPRE: crop yield prediction and allocation using linear regression and greedy algorithm

Date

2025

Adviser

John Cedric C. Gaza

Principal

Mabel S. Buela

Abstract

The Department of Agriculture (DA) promotes polycropping for sustainability. Crop rotation, a polycropping approach for sustainable farming, combines the traits of different crops to restore and nourish the soil it is planted on. Different plots have unique factors that affect crop yield and many researchers have proposed a prediction model for this problem and have generated great results. Nevertheless, it is imperative to test out more models to find one more interpretable and accurate than the previous. In this paper, two machine learning models were employed: KNN and linear regression. With the data gathered from an open source database and an institution, the researchers developed a system which consists of two parts. First, a machine learning algorithm is used to forecast the yield of crops and rank them based on their compatibility to the plot with regards to its soil quality. This is then inputted to a simple greedy algorithm, along with the previously planted crops on that plot, to output a crop allocation of the highest profiting crops with regards to the cropping rotation constraints. Yield forecast for the real-world data from the institution returned an average MAPE of 54.5% and 70% for KNN and linear regression respectively, this means that at its current stage the model is highly inaccurate with its prediction. However, further development of this system can allow for its use for simple forecasting, and will serve as the foundation for crop scheduling. The researchers recommend forecasting the market rather than treating it as a stagnant variable, and considering time as a feature.

Language

English

Location

UP Rural High School

Notes

To access this capstone, please contact the UP Rural High School Library at uprhslibrary.uplb@up.edu.ph. You may also visit the library in person, provided you secure prior confirmation from the librarian. We will be happy to assist you.

Document Type

Capstone

This document is currently not available here.

Share

COinS