Description:MACHINE LEARNING BASED PLANT SELECTION PREDICTION FRAMEWORK USING INTERNET OF THING

Field of Invention
The present invention relates to all the farmers to predict the suitable crop in their farmland by applying machine learning through IOT devices. By introducing this type of device, we can reduce the ambiguity in the farmers in growing crops and also improve the yield.
The objectives of this invention
The objective of the crop recommendation system using a machine learning algorithm is to provide farmers with optimized crop recommendations based on the analysis of various factors such as soil type, weather conditions, and historical crop yields. The system uses a machine learning algorithm to process and analyze large amounts of data and provide accurate and personalized crop recommendations, leading to improved crop yields, reduced costs, and increased overall efficiency in agriculture.
Background of the invention
In (WO2016/2017191872A1), the technique and framework for recognizing a plant abnormality have been released. It involves learning from a variety of model predictions with various characteristics based on real-time plant data gathered to produce the highest-accurate prediction value, followed by accurately identifying a problem to detect a plant deviation and giving an early warning. In (US2018/11263707B2), in order to forecast the growth of crops and select a set of agricultural procedures that, if carried out, would optimise crop yield, a crop forecasting system executes a variety of machine learning processes. In (US2015/10349484B2), A plant circumstances prediction is made using the at least one test input and an algorithm for prediction, which is based on a set of training variables with at least a single training data and at least one training results, each of which matches to a training input.
In (CN2019/111026200A), The innovation is associated with the field of cultivation and reveals an Internet of things for the prediction, early detection, as well as management of agricultural pests and growth conditions. By using agriculture diseases, pests of insects, as well as development circumstances, this innovation also provides a prediction, preventive, and control technique for the Internet of Things. In (US2017/0161560A1), a framework and procedure for forecasting crop yield. The process involves obtaining information about monitoring for at least one crop, that consists of at least one multidimensional component representing the crop; analysing the digital multimedia component using machine vision; and retrieving the information based on the results of the analysis.
The crop recommendation system using a machine learning algorithm is a tool designed to help farmers make informed decisions about the type of crops to grow on their farms. This system is built on the idea that machine learning algorithms can be used to predict the performance of different crops based on various factors such as soil type, weather patterns, and previous crop performance.
Traditionally, farmers have relied on their own experiences and knowledge to decide what crops to grow. However, this approach can be limited and often results in low crop yields, which can negatively impact the farmer's income and ability to provide food for their community. This information can then be used by farmers to make informed decisions about what crops to grow, which can lead to improved crop yields and overall profitability. Additionally, the system can also help farmers identify the best practices for growing their crops, reducing the risk of crop failures and improving the sustainability of their operations.

Summary of the invention
The crop recommendation system using a machine learning algorithm is an innovative solution that helps farmers in making informed decisions about what crops to grow and when to grow them. It leverages the power of machine learning algorithms to analyze vast amounts of data and generate crop recommendations based on specific geographic and climatic conditions. This system takes into consideration factors such as soil type, weather patterns, and precipitation levels, as well as other relevant data, to provide farmers with accurate crop recommendations tailored to their specific needs. The system can be easily integrated into existing farming operations, providing farmers with a simple and effective tool to optimize their crop selection and production processes. With its ability to analyze large amounts of data and generate accurate recommendations, the crop recommendation system using a machine learning algorithm has the potential to revolutionize the way farmers approach crop production, helping them to optimize yields and reduce costs.
Detailed description of the invention
The proposed system tries to minimize the drawbacks of the existing system. It considers N, P, K levels, pH, soil moisture, temperature, and humidity. Since all the vital parameters are being considered, the resultant system will be more accurate. It is cheaper and easy to set up. Replacement is also easy. The parameter transmission delay is reduced. It does not need much technical skill to set up. It uses Random Forest classifier which reduces the problem of high variance and offers high accuracy. By determining various trees during the training and testing process of the ML model. The supervised learning for splitting, association, regression, and other assignments uses random decision forest classifiers as a study tool. After that, the class that best indicates the manner of decision trees' categorization or predictive regression is determined. Individual trees can be better classified using the random choice forest classifiers than groups of samples used for training and testing. The algorithm connects an input to an output in supervised learning. The term alludes to the idea that the learning principles plant many different choice trees in a forest. The kaggle data set is generally used to analyze the suitable crop by processing data with Random Forest Classifier.
It is a data set consisting of predefined attributes like NPK values, humidity, temperature, ph, soil moisture level. This data set to train and test the model by using the random forest classifier and regression technique. The DHT11 sensor is used to find the temperature and Humidity. This sensor has a separate NTC regarding temperature calculation and a microcontroller of 8-bits for regular data. This sensor has a minimalistic mechanism to connect to other microcontrollers. Sensor measures values of temperature upto 50°C and humidity values ranging from 20% to 90%. Electrical conductivity of the earth surface and the soil resistance will be detected by this sensor. When execution of the threshold output happens, it gives the digital output. The water’s volumetric content is measured from the probes of this sensor. These two probes pass the flow of current through the clay, and moisture level resistance. Soil’s electricity is directly proportional to the conductivity of the soil nature. Due to this the dry soil will have less conductivity compared to the wet soil. This will practically be measured by soil moisture sensor. The architecture of the system resembles the inflow and outflow of the data. By using the DTH11 sensor we measure the humidity and temperature of the soil. In addition, with soil moisture level by using the soil moisture sensor. These inputs straight away to the Arduino UNO board. With these inputs and machine learning algorithms, we train the model and inclusion of the nitrogen phosphorus, and potassium content in the air. This will be analyzed and predicted under Random Forest Algorithm and regression analysis. By using the Django framework we have developed a web application that is used for the interaction of the user and the machine learning model
The crop recommendation system using a machine learning algorithm is an innovative solution for farmers to optimize their crop yield and improve the quality of the crops. This system utilizes machine learning algorithms to analyze the data related to various crop growing conditions such as soil quality, weather patterns, and fertilizer application to provide farmers with customized recommendations for their crops.
The system collects data from various sources such as satellite imagery, weather stations, and soil sensors. This data is then analyzed by machine learning algorithms to identify the factors that influence crop growth. Based on this analysis, the system develops models that predict the growth and yield of crops based on different combinations of these factors.
Once the models are developed, farmers can input the conditions of their fields into the system, including the soil quality, the weather patterns, and the fertilizer application. The system then uses these inputs to generate customized crop recommendations based on the most optimal conditions for each crop.
The system also provides recommendations for fertilization and irrigation to help farmers maintain optimal soil conditions for their crops. This includes recommendations for the application of organic or inorganic fertilizers and the frequency of irrigation. The system can also be used to monitor crop growth and yield over time, providing farmers with real-time feedback on the impact of their decisions.
The crop recommendation system is an easy-to-use tool that can significantly improve crop yield and quality. By using machine learning algorithms to analyze large amounts of data, the system provides farmers with customized and scientifically-based recommendations that are tailored to their specific needs. This system has the potential to revolutionize the way farmers grow crops, resulting in increased productivity and profitability.
The system shall use a Random Forest method for the processing of decision trees according to parametric characteristics of crop forecasts and Real Time Data values supplied by integrated devices such as soil moisture and Ambient Temperatures. A user-friendly web application making it possible for farmers to store and input soil data is available through the created system. Before recommending the best crops for that soil, the program shall display temperature and moisture levels on the land surface as well as current climatic conditions. Using the account management and crop selection history features of this web application, farmers will be able to monitor the development of their decisions on selecting land and crops.
The system's precision was tested in a real-time setting, and the findings showed a high level of accuracy with a notable improvement in the choice of acceptable crops. Additionally, the system readily interfaces with other Internet of Things (IoT) devices, enabling additional data collecting for more accurate crop forecast analysis and planning.

4 Claims & 1 Figure
Brief description of Drawing
In the figure which illustrates exemplary embodiments of the invention.
Figure 1, Architecture of the proposed plant selection model. , Claims:The scope of the invention is defined by the following claims:

Claim:
1. A system/method for prediction and selection of plants using machine learning based algorithm and pre-defined data set, said system/method comprising the steps of:
a) The system collects the temperature and humidity values (1) from the sensors and sent to the IOT module (2).
b) Then Machine Learning module (3) exploring the data from the data set and combining it with the suitable algorithm.
c) This device renders on web (4) with the help of the model generated by the pickle package (5) where the prediction is done based on the past data. Predict the suitable plant growing on the land (6, 7).
2. As mentioned in claim 1, the IOT module collects the requirements from the hardware devices and is used for the prediction of the data.
3. According to claim 1, the soil moisture level, temperature and humidity levels will be directed to the Arduino uno board.
4. As per claim 1, training and testing of the model will be handled by the machine learning module and data will be predicted accurately.