ML and IoT-based Soil Type Analysis and Prediction System

The agricultural economic sector looks to be dominant in India’s social structure. Sometimes farmers sow the crop without knowing the soil type and with more fertilizers to satisfy the customer preferences. This will affect productivity and decrease the nutrients present in the soil. It is important to identify the soil type, so then the farmer will choose the crop rightly. The solution will be done by employing new technologies like the Internet of Things (IoT) and Machine Learning (ML). In this work, the IoT is used to capture and send the details of the soil type and its location. Then ML is used to identify the soil type. To analyse and anticipate the kind of soil, data will be collected from Kaggle, pre-processing is done secondly, feature extraction is being used to extract important attributes from the image. The Wavelet Transform is employed as a feature extraction technique. The extracted data is subsequently employed in training and testing the ML model to classify the kind of soil. Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA) algorithms are chosen as an ML model. Both models are evaluated using the metrics and identified that the SVM will be best for soil classification. Finally, the mobile app is created with a simple Graphical User Interface (GUI), so farmers can easily access the proposed technique. The SVM is employed in the mobile app to predict the soil type based on the information acquired through IoT.