Comparison Analysis of Oil Crop Yield Prediction in Magway Region using Machine Learning Methods
- 1. Comparison Analysis of Oil Crop Yield Prediction in Magway Region Using Machine Learning Method Presented by Ei Ei Moe Tun Associate Professor University of Computer Studies, Mandalay
- 2. 2 Outlines • Abstract • Objectives of the Propose Works • Introduction • Background Theory • Proposed System Architecture • Data Preprocessing • Data Classification with Three Methods • Compare Classification Accuracy • Conclusion • Future Works • References
- 3. 3 Abstract • In Myanmar economy, agriculture is an important factor for farmer. Today trend is going toward the smart farming for agriculture all over the world. • Machine Learning is a fundamental methodology for accomplishing pragmatic and powerful answers to this issue. Machine learning is an important decision support tool for crop yield prediction, including supporting decisions on what crops to grow and what to do during the growing season of the crops. • Oil Crop Yield Prediction includes foreseeing the yield of the oil crop from accessible chronicled accessible information like climate boundary, soil boundary, and noteworthy harvest yield. • This paper focuses on foreseeing the yield of the harvest depends on the current information by utilizing three machine learning algorithm. • Real data of the Magway region were utilized for building the models and the models were tested with samples.
- 4. 4 Abstract (con’t) • This is accomplished by applying three machine learning techniques, for example, Neural Networks, Support Vector Machines, and Decision Tree strategies for oil crop information in the Magway region. • Oil crops in the Magway district are sesame, groundnut, and some sunflower. • The exploration is expected to assist farmers in predicting the yield of the crop before developing onto the agribusiness field. • The proposed framework is intended to arrive at smart farming for Myanmar agriculture. • The prediction will assist the rancher in predicting the yield of the crop before developing onto the agriculture field. • To anticipate the crop yield in future precisely machine learning tools, a generally incredible and well known managed ML algorithm is utilized.
- 5. 5 Objectives of the Propose Works •To achieve maximum oil crops (sesame, groundnut and some sunflower) yields and effective fertilizer usage for next five year based on average parameters like rainfall, temperature, fertilizers, pesticides, ph level, and other atmospheric conditions and parameters in Magway region. •To alert the farmer for the minimum crops yields is caused by over usage of fertilizers. •To assist the farmer and agronomist who use the wrong usage of fertilizers that may cause the health hazard.
- 6. 6 Introduction • Machine learning, which is a branch of Artificial Intelligence (AI) focusing on learning, is a practical approach that can provide better yield prediction based on several features. • Machine learning (ML) can determine patterns and correlations and discover knowledge from datasets. • The models need to be trained using datasets, where the outcomes are represented based on past experience. • The predictive model is built using several features, and as such, parameters of the models are determined using historical data during the training phase. • For the testing phase, part of the historical data that has not been used for training is used for the performance evaluation purpose. • Prediction is the methodology to predict the yield of the crops using different parameters like rainfall, temperature, fertilizers, pesticides, ph level, and other atmospheric conditions and parameters. • This paper is used to monitor and predict crop yields in remote areas and cities around Magway region.
- 7. 7 Background Theory • Artificial Neural Network (ANN) • Decision Tree • Support Vector Machine
- 8. 8 Artificial Neural Network (ANN) • ANNs are versatile machine learning algorithms that can perform classification and regression tasks, and even multivariate output tasks. • The model of a single artificial neuron can be understood in very similar terms to the biological model. • The figure 1 shows the multilayer neural network with one input layer, one hidden layer, and one output layer. • Activation functions are used to get precise output. • By utilizing the ANN algorithm for oil crop yield helps to ensure the precision and effectiveness of crop yield for maximizing crop production. Fig 1. Multilayer Neural Network
- 9. 9 Decision Tree • Decision trees are used in data mining and machine learning. • Decision trees are the most popular machine learning algorithm because of their intelligence and simplicity. • It is as often as possible utilized by the scientists to characterize the information. • This work deals with the prediction of crops yield as high yield or low yield using Decision tree algorithm. • The result of decision tree was compared with neural network and support vector machine classifier. Fig 2. A simple decision tree
- 10. 10 Support Vector Machine • Support Vector Machines (SVMs) are a particularly efficient and flexible type of computation. • A support vector machine (SVM) is a relatively simple machine learning algorithm used for classification. • They are used not only for classification but also for prediction. • Basically, SVM finds a hyperplane that creates a boundary between data types. • The SVN algorithm is used for oil crop yield helps to ensure crop yield for maximizing crop production. Fig 3. The Model of the Support Vector Machine
- 12. 12 Overview of Data set • This research utilizes crop yield prediction strategies to forecast the appropriate crop at the remote areas and cities around Magway region. Magway region has five districts. Magway district, Minbu district, Pakokku district, Thayet district and Gangaw district. • The oil crop data are collected around these districts. Nine year historical data of Magway region is used to predict the oil crop yields and to help for farmer for usage of fertilizer and health hazard. • Big amount of oil crop data in Magway region is recorded because of the five number of districts, the data elements in this districts and the observations are recorded yearly. • Data were collected from 2010-2019 i.e. nine years are collected to analyze the oil crops prediction in Magway region. • These heterogeneous sources are stored the data as different format and different unit i.e. one district stored the unit of temperature attribute as kelvin and other as C.. • These types of different format and different unit are integrated with same form and format at the preprocessing step of the proposed system.
- 13. 13 DATASE DESCRIPTION Variable Description Year The year 2010-2019 of Magway districts oil crop data Township 25 townships of Magway Division District Five districts in Magway Division Crops Plant like a sesame, peanut, sunflower Area Agriculture plants region total area Temperature Average temperature of each township in a year Rainfall Average rainfall of each township in a year Humidity Average humidity of each township in a year Production The crop’s production in Tons Yield Average yield per acre Yield Class Excellent (E) very good (VG) good (G) poor (P). TABLE I. DATASE DESCRIPTION
- 14. 14 SAMPLE DATASET TABLE II SAMPLE DATASET
- 15. 15 Data Preprocessing • In the pre-process, the data which is collected from the five districts of Magway region. There are four step in data preprocessing step. • In the first step, noise values are removed by using noise detection algorithm. The data needs to be cleaned at this stage because the raw data contains missing values. • In the second step, repeated values are removed by estimating the similar values in the stored data. • Data canonicalization is implemented in the third step. • The last fourth stage, the components are analyzed to understand the nutrients in the name of fertilizer. This step is used to recommend the farmer and agronomist.
- 16. 16 Data Classification with Three Methods • The output of the data preprocessing step enter the data classification as input for classify three classes using three machine learning methods. • These three methods is used to predict the yield and classify the four class of crop yield. • To model the relationship between the conditional attributes and decision attribute used in crop yields prediction and classification of yields class will use three machine learning methods such as Neural Network, Decision Tree, and SVM. • The implementation of the research is used R programming. In the future, we predict the yields of oil crops in Magway region with the available attributes and calculate the accuracy with the detail data of five districts at the Magway region. • And then, we want to recommend the farmers for the usage of fertilizer and health hazard with the classified result of three models.
- 17. 17 Compare Classification Accuracy • The crop data set with 500 instances in three different machine learning methods was run. • The accuracy result is analyzed with 80% of data set as training data and the rest 20% is testing data. • The analysis results are shown in Table 3. • Among these methods, the SVM method is the worst accuracy over the data set with train 80% and test 20%. Sometimes, by using the training and testing is changed the accuracy of the SVM is better than the other two methods. • Finally, the system concluded that the overall good accuracy depends on the more training data. • Figure 4 shows the comparison result of the three methods by using bar chart with training 80% and testing 20% with 500 instances of crops dataset.
- 18. 18 Compare Classification Accuracy(con’t) ML Method Precision Recall F-Measure Decision Tree 1.0 1.0 1.0 SVM 0.888 0.880 0.887 ANN 1.0 1.0 1.0 TABLE III. ACCURACY RESULTS BY USING THREE METHODS WITH TRAINING 80% AND TESTING 20% IN CROPS INSTANCE 500
- 19. 19 Sample Comparison of Three Methods using Bar Chart 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1 1.02 Decision Tree SVM ANN Comparison of Three Methods Precision Recall F-Measure Fig 4. Sample Comparison of Three Methods for Training 80 % and Testing 20%Performance Evaluation with 500 instances
- 20. 20 Compare Classification Accuracy(con’t) • And then, the crop data set with 1000 instance in three different machine learning methods was run. • The accuracy test result is analyzed with 70% of data set as training data and the rest 30% is testing data. • The analysis results are shown in Table 4. • Among these methods, the SVM method and ANN is the best accuracy over the data set with train 70% and test 30%. • Table 5 shows the analysis results of crops data set with 1000 instances in training 80% and testing 20%. • In this evaluation, the overall f-measure is the best for ANN and SVN methods. • Figure 5 shows the comparison result of the three methods by using line graph with training 70% and testing 30% with 1000 instances of crops dataset.
- 21. 21 Compare Classification Accuracy(con’t) ML Method Precision Recall F-Measure Decision Tree 0.887 0.88 0.88 SVM 1.0 1.0 1.0 ANN 1.0 1.0 1.0 TABLE IV. ACCURACY RESULTS BY USING THREE METHODS WITH TRAINING 70% AND TESTING 30% IN CROPS INSTANCE 1000
- 22. 22 Compare Classification Accuracy(con’t) ML Method Precision Recall F-Measure Decision Tree 0.887 0.88 0.88 SVM 1.0 1.0 1.0 ANN 0.90 0.99 0.98 TABLE V. ACCURACY RESULTS BY USING THREE METHODS WITH TRAINING 80% AND TESTING 20% IN CROPS INSTANCE 1000
- 23. 23 Sample Comparison of Three Methods using Line Chart 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1 1.02 Decision Tree SVM ANN Comparison of Three Methods Precision Recall F-Measure Fig 5. Sample Comparison of Three Methods for Training 70 % and Testing 30% Performance Evaluation with 1000 instances
- 24. 24 Conclusion • In this paper, a comparison for Neural Networks, Support Vector Machines, and Decision Tree Classifier are shown for analysis of oil crop information in the Magway region. • This paper analyzes the crop yield production based on available oil crop data. The ML was used to predict the crop yield for maximizing crop productivity. • According to the resulting factors, the performance of Neural Networks and Decision Tree is more accuracy and more precision than SVM Classifier for training 80% and testing 20%. • But the performance of ANN and SVM is more accuracy and more precision than Decision Tree. for training 70% and testing 30% in 1000 oil crops instances. • Finally, the system concluded that the overall good accuracy depends on the more training data. Crop yield prediction is an important issue for farmers and agronomists.
- 25. 25 Future Work • This paper means to propose and actualize an information framework to estimate the crop yield creation from the pool of gathered past information. • In the future, this model would be implemented with much more efficient technique such as IOT based smart farming system with more efficient data set. • And then, we want to manage the usage of fertilizer for farmer health.
- 26. 26 REFERENCES • Teresa Priyanka, Pratishtha Soni, C. Malathy, Recureeived 05 November 2018 Revised: 23 November 2018 Accepted: 02 December 2018 “Agricultural Crop Yield Prediction Using Artificial Intelligence and Satellite Imagery”.J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed., vol. 2. Oxford: Clarendon, 1892, pp.68–73. • Prashant Govardhan, Rasika Korde, Rashi Lanjewar (December 2018) “Survey on Crop Yield Prediction Using Data Mining Techniques”. • Eissa Alreshidi1 (No. 5, 2019) “Smart Sustainable Agriculture (SSA) Solution Underpinned by Internet of Things (IoT) and Artificial Intelligence (AI)” • Mona Kadryb , Ayman E. Khedra , Ghada Walidb , International Conference on Communication, Management and Information Technology (ICCMIT 2015) , “ Proposed framework for implementing data mining techniques to enhance decisions in agriculture sector Applied case on Food Security Information Center Ministry of Agriculture, Egypt”. • Prof. D.S. Zingade1, Omkar Buchade2, Nilesh Mehta3, Shubham Ghodekar4, Chandan Mehta5 (Volume 4, Special Issue 5, Dec.-2017) “Crop Prediction System using Machine Learning” • Zainab Arshad, Sohail Masood Bhatti, Huma Tauseef and Arfan Jaffar, “Heart Sound Analysis for Abnormality Detection”, Intelligent Automation & Soft Computing DOI:10.32604/iasc.2022.022160, Lahore College for Women University, Lahore, Pakistan corresponding Author: Sohail Masood Bhatti, 16 September 2021. • www.coursehero.com • www.archive.org • www. towardsdatascience.com • www. medium.com
- 27. 27 Thanks for Attending my presentation