Machine Learning-based Crop Recommendation System in Biswanath District of Assam

Author: Samikhya Bhuyan, Dilip Kumar Patgiri, Simanta Jyoti Medhi, Roopali Patel and Tabuiliu Abonmai

Download PDF

Abstract

The present study makes attempt to formulate a crop recommendation system based on soil physical property. The study was undertaken in the Biswanath district of Assam which is situated 26°35' to 27°00' of latitude and 92°50' to 93°50' of longitude, with a total area of 14,15,000 ha. The 180 soil samples were collected from different location and physical properties were analyzed in laboratory using standard protocol. The selected physical properties includes soil texture, bulk density, particle density, total porosity, hydraulic conductivity, maximum water holding capacity, volume expansion, field capacity, permanent wilting point, available water, microaggregate, macroaggregate and mean weight diameter. The coefficient of variation (CV) was used for interpretation of variability of the soil properties. Silt, clay, hydraulic conductivity, were identified to be the most variable soil indicators (CV > 35%). Available water was found as the moderately variable parameters (CV 15–35%). The least variation (CV < 15) were found in bulk density, particle density, total porosity, field capacity, mean weight diameter, microaggregate, macroaggregte and volume expansion. The results of the physical properties were fit in the machine learning model and a farmer’s friendly crop recommendation system was developed.

Keywords

Crop recommendation system, Physical property, coefficient of variation, Mean weight

Conclusion

In Assam's Biswanath District, crop recommendation systems was forecasted using machine learning techniques. The crop recommendation system utilized application of the “decision tree classifier model”. The model's accuracy was almost 94%. A web application is being created to link the model. This application presents the model to the user. The study recommends that farmers can change the type of crop planted in their field during various season depending on the physical properties of the soil in order to minimize crop failure.

References

-

How to cite this article

Samikhya Bhuyan, Dilip Kumar Patgiri, Simanta Jyoti Medhi, Roopali Patel and Tabuiliu Abonmai (2023). Machine Learning-based Crop Recommendation System in Biswanath District of Assam. Biological Forum – An International Journal, 15(3): 417-421.