Article Preview
TopIntroduction
The rapidly evolving technology shows its relevance in the field of agriculture. Agricultural production forecasting is a scientific method for estimating crop yield before harvest (Liakos et al., 2018). Regardless of weather and market conditions, research on plantation crop productivity is crucial to strengthen the stability of farmers' financial situations year-round (Vaidya & Katkar, 2022). Building agricultural risk management and prognostic reasoning models is essential for retaining current farmers and attracting new ones due to the various barriers in the agriculture field (Krishna et al., 2022). A predictive reasoning model that protects farmers against agricultural risks is the goal of this research. This study proposes a paradigm for supporting scientific decision-making in agriculture.
The research focuses on the areca nut crop grown in the Mangaluru region of Karnataka, India, and real data is collected for the study from the farmers. Arecanut crop is influenced by the different agronomic elements, namely land type, plantation type, areca nut variety, sunlight rate, shading, water availability, irrigation type, fertilizer usage, and pesticide usage.
Farmers may effectively plan the plantation process using this model. Farmers can find the best combinations of features impacting crop productivity to avoid loss. It is a scientific model that gives farmers appropriate plantation methods based on evolving agronomic factors. Crop yield prediction using existing models is only possible through statistical analysis (Majumdar et al., 2017). Consequently, there is a significant demand for scientific reporting and analysis. The suggested model is built using data analytics techniques that facilitate the farmers to make knowledge-driven proactive decisions. Data analytics mainly involves data collection, maintenance, analysis, and prediction using statistical-driven data mining techniques (García-Peñalvo et al., 2021).
The construction of any predictive model involves two phases: data modeling and prognostic reasoning (predictive analysis). The data modeling phase involves data collection, preprocessing, and feature selection tasks (Janssen et al., 2017). The various feature selection methods are available and used to select appropriate features before predictive analysis (Jose et al., 2019). But these methods do not guarantee uniformity and relevance between the features. Thus, this research uses a formal statistical test called the chi-square test to measure the relevance between the features and discover the most influential features.
Based on historical data, prognostic reasoning helps to analyze the several features that influence crop productivity (Diriba & Borena, 2013). Prognostic reasoning is performed using various data analytics tasks such as classification, knowledge discovery, and decision-making (Pakkala et al., 2021). The different classification algorithms' performance is measured based on various statistical factors, and finally, discovered knowledge is presented to the farmer.
The primary objectives of this research are as follows:
- •
Determination of survival rate of newly planted areca nut palm.
- •
Estimation of flowering time of areca nut tree concerning areca nut variety.
- •
Classification and prediction of crop yield.
Individual farmers can use the proposed scheme for risk management in agriculture and obtain early knowledge on the plantation strategy of the areca nut crop. Also, this model can be utilized by any agriculture department for scientific analysis and reporting.
The rest of the paper is structured as follows: We discuss some past related work on agriculture predictive models in section 2. The proposed methodology is detailed in section 3, along with the feature selection algorithm. Section 4 discusses the experimental findings. Section 5 summarises the conclusion and future work.
TopThe review and discussion focus on the agricultural predictive modeling and feature importance literature employing clustering, classification, and decision making.