Wheat Yield Prediction using Temporal Fusion Transformers

Title

Wheat Yield Prediction using Temporal Fusion Transformers

Creator

Junankar T.; Sondhi J.K.; Nair A.M.

Description

In precision framing, Machine Learning models are an essential decision-making tool for crop yield prediction. They aid farmers with decisions like which crop to grow and when to grow certain crops during the sowing season. Many Machine Learning algorithms have been used to support agriculture yield prediction research, but it is observed that Deep Learning models outperform the benchmark Machine Learning algorithms with a significant difference in accuracy. However, though these Deep Learning models perform better, they are not preferred or widely used in place of Machine Learning models. This is because Deep Learning methods are black box methods and are not interpretable, i.e., they fail to explain the magnitude of the impact of the features on the output, and this is unsuitable for our use case.In this paper, we propose using Temporal Fusion Transformer (TFT), a novel approach published by Google researchers for wheat yield prediction viewed as a Time Series Forecasting problem statement. TFT is the state-of-the-art attention-based Deep Learning architecture, which combines high-performance forecasting with interpretable insights and feature importance. We have used TFT to perform wheat yield prediction and compare its performance with various Machine Learning and Deep Learning algorithms. 2023 IEEE.

Source

2023 2nd International Conference for Innovation in Technology, INOCON 2023

Date

2023-01-01

Publisher

Institute of Electrical and Electronics Engineers Inc.

Subject

Crop Yield Prediction; Deep Learning; Machine Learning; Temporal Fusion Transformers; Time Series Forecasting

Coverage

Junankar T., CHRIST(Deemed to Be University), Department of Data Science, Pune, Lavasa, India; Sondhi J.K., CHRIST(Deemed to Be University), Department of Data Science, Pune, Lavasa, India; Nair A.M., CHRIST(Deemed to Be University), Department of Data Science, Pune, Lavasa, India

Rights

Restricted Access

Relation

ISBN: 979-835032092-3

Format

Online

Language

English

Type

Conference paper

Identifier

https://doi.org/10.1109/INOCON57975.2023.10101144

https://www.scopus.com/inward/record.uri?eid=2-s2.0-85157974107&doi=10.1109%2fINOCON57975.2023.10101144&partnerID=40&md5=f711c3f13b878b46d98e74dcc0dd9c83