International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395 -0056
Volume: 04 Issue: 01 | Jan -2017
p-ISSN: 2395-0072
www.irjet.net
An Empirical Study on Mushroom Disease Diagnosis: A Data Mining Approach Dr. Dilip Roy Chowdhury1, Subhashis Ojha2 1Assistant
Professor, Dept. of Computer Science & Application, University of North Bengal, West Bengal, India 2Assistant
Teacher, Mathurapur B.S.S. High School, Malda, West Bengal, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In this paper, a data mining application is
introduced for selecting highly effective factors or symptom of different disease diagnosis in mushroom yield. The study also focuses on several factors causing a specific mushroom disease. Highly potential symptoms among several factors were focused out for better management in this regard. That’s why data mining techniques are being used for ranking among symptoms. This paper focuses on identifying specific diseases among several diseases using a data mining classification based approaches. Real data has been taken from mushroom farm and thereafter purification of potential factors is done through data mining approaches. The classification technique and disease prediction of mushroom dataset were prepared using Naïve Bayes, SMO and RIDOR algorithms. A statistical comparison has been produced in order to find the best symptoms needed for mushroom disease diagnosis. Besides this, it searches the best performing classification algorithm among all. Key Words: Classification Algorithm, Data Mining, Mushroom, Ripple Down Rules (RIDOR), Sequential Minimal Optimization (SMO).
1. INTRODUCTION Mushroom is one type of fungus type plant containing no chlorophyll. There is around 45000 type fungus available in the world. Among them, around 2000 fungus are edible vegetable food. Mushroom can be edible and non-edible. Cultivating mushroom in scientific ways reduces the probability to occur poison in mushroom yield. In our country, four kinds of mushroom are available namely Button Mushroom, Oyster Mushroom, Paddy Straw Mushroom, Milky Mushroom. Like all other crops, a huge number of biotic and abiotic agents or factors affect mushroom yields. Among the biotic agents, fungi, bacteria, viruses, nematodes, insects and mites are responsible for damaging in mushrooms directly or indirectly. A number of harmful fungi are seen in compost and casing soil during the cultivation of mushroom. Many of these work as competitor molds thereby adversely affecting spawn run. On the other hand the fruit body at various stages of crop growth is hampered by others, thus producing distinct disease symptoms. If this continues, total crop failures occur based © 2017, IRJET
|
Impact Factor value: 5.181
|
on the stage of infection, quality of compost and environmental conditions. At any phase of growth an undesirable growth or development of certain molds can occur and can adversely affect the final mushroom yield [1]. Finding the diseases and its reason in context to mushroom is very much necessary where there is a scarcity of domain expert on this field. Data mining is a process to get the meaningful data from the large data scattered in large data repository. By using different tools and techniques, right information is provided for data mining process. It is popularly known as information or knowledge discovery, which is one of the recent trends found to be useful in several complex fields. It is the process of evaluating data from different outlooks and summarizing it into useful information that can be used to identify the symptom of different diseases in mushroom. Using data mining mushroom data set can be analyzed. Data mining allows users to investigate data from many different dimensions or angles, categorize it, and summarize the relationships identified [2]. Technically, data mining is used for set up relationship among several fields in dataset. The objective of application of data mining techniques on mushroom dataset is to analyze such data and to find relative importance of different disease parameters for differential diagnosis in mushroom. A specific disease diagnosis is not a result of one deciding factor, in addition it heavily hinges on multiple symptoms. This paper identifies the symptoms associated with mushroom disease that help to the farmers/growers to improve the quality of production as well as to cope up the huge losses for damaging. This study reveals the accuracy of some classification techniques has been measured and relative importance has been found among several disease symptoms of mushroom. The main objective of this proposed work is to find out differential diseases diagnosis and identification of the highly importance attribute for disease diagnosis and also to find the optimal classification mechanism.
2. LITERATURE REVIEW Data mining applications are being recently used in agricultural research and many activities can be observed in ISO 9001:2008 Certified Journal
|
Page 529