Agriculture is one of the foremost and the minimum salaried employment in India. Data mining be able to fetch an explosion in the agriculture field by altering the profits scenario through growing the optimum crop with crop yield prediction, which is a difficult task because of the climatic factors, soil fertility, nutrients and so on. Precise crop forecast requires fundamental understanding of the functional association between crop and input parameters and to predict the crop yield in advance we developed an Adaptive Lemuria algorithm. Our proposed model comprises of Deep Belief Network for feature learning and pre-training, Decision tree & K-Means clustering (HDTKM) with Particle Swarm Optimization (PSO) for training to attaining global solution and Naive bayes clustering with PSO for testing to get optimum result. The forecast made by our proposed algorithms will aid the ranchers to choose which crop to cultivate to get the extreme yield. The experimentation was conducted to verify the performance of our proposed framework in python with Anaconda Spyder and outcome attains 98.35 % of accuracy with an error rate of 0.0314, which is relatively higher than the existing methodologies.

农业是印度最重要也是最低工资的工作之一。数据挖掘能够通过种植具有作物产量预测的最佳作物来改变利润场景，从而在农业领域取得爆炸性增长，这是一项艰巨的任务，因为气候因素、土壤肥力、养分等。精确的作物预测需要对作物和输入参数之间的功能关联有基本的了解，并且为了提前预测作物产量，我们开发了一种自适应 Lemuria 算法。我们提出的模型包括用于特征学习和预训练的深度信念网络、带有粒子群优化 (PSO) 的决策树和 K 均值聚类 (HDTKM) 用于训练以获得全局解决方案，以及用于测试以获得最优解的朴素贝叶斯聚类与 PSO结果。我们提出的算法所做的预测将帮助牧场主选择种植哪种作物以获得极高的产量。进行实验是为了验证我们提出的框架在 python 中使用 Anaconda Spyder 的性能，结果达到 98.35% 的准确率，错误率为 0.0314，这相对高于现有方法。