Enhancing Crops Production Based on Environmental Status Using Machine Learning Techniques

The suitable crop for a particular location is necessary for agriculture to bring the most productivity. Here we have designed a model that contains prediction and recommendation with machine learning approaches that determines productivity based on the parameters humidity, rainfall, and temperature. For the prediction, we have applied k-nearest neighbor (KNN), support vector machines (SVM), random forest, naive Bayes' classifier and logistic regression, collaborative filtering, and Multi-Condition Filtering algorithms. After training the dataset and applying these algorithms, we have made a comparison of the algorithms by analyzing the precision. On the other hand, for the recommendation, we have applied collaborative filtering and Multi-Condition Filtering algorithms where these algorithms take input parameters. And then, in the collaborative filtering the input parameter is compared with the trained data that is already in the system and filters out the best 5 crops as output based on cosine similarity and Multi-Condition Filtering algorithm categorizes the crop with a different combination of the high, low and moderate ranges of the input parameter and displays the crop accordingly.