Data mining in agriculture

Data mining in agriculture is a research topic that applies data mining and data science techniques to the agricultural sector. Recent advancement in technology have made it possible to collect large amount of data related to agricultural activities, which can then be analyzed to extract valuable information.

Applications

Relationship between sprays and fruit defects

Fruit defects are often recorded for various reasons, including insurance reasons when exporting fruit overseas. This can be done manually or through computer vision which detects surface defects when grading the fruit. Spray diaries are a legal requirement in many countries typically recording the date of application and the product used. It is known that spraying can cause various defects in different types of fruit. Fungicidal sprays are often used to prevent rots from developing on fruits. It is also known that some sprays can cause russeting on apples. Currently, much of this knowledge is based on anecdotal evidence; however, there have been efforts to explore the use of data mining in horticulture.

Prediction of problematic wine fermentations

The fermentation process of wine impacts the productivity of wine-related industries as well as the quality of the wine. Data science techniques, such as the k-means algorithm, and classification techniques based on the concept of biclustering have been used to study the process of fermentation in order to predict problematic wine fermentations. These methods differ from techniques used to classify different kinds of wine. See the wiki page Classification of Wine for more details.

Predicting metabolizable energy of poultry feed using group method of data handling-type neural network

A Group Method of Data Handling(GMDH)-type network combined with an evolutionary method of genetic algorithm, was used to predict the metabolizable energy of feather meal and poultry offal meal based on their protein, fat, and ash content. Data samples from published literature were collected and used to train a GMDH-type network model. The novel approach of combining GMDH-type network with an evolutionary method of genetic algorithm can be used to predict the metabolizable energy of poultry feed samples based on their chemical content. It is also reported that the GMDH-type network can accurately estimate the poultry performance from their dietary nutrients such as metabolizable energy, protein and amino acids.

Detection of diseases from animal sounds

The detection of diseases on farms can positively impact the productivity of the farm by reducing contamination to other animals. Moreover, the early detection of the diseases can allow the farmer to treat and isolate the affected animal as soon as the symptoms appear. Sounds emitted by pigs, such as coughing, can be analyzed for disease detection. A computational system is currently being development to monitor pig sounds through microphones installed in the farm, and which is also able to differentiate between the various sounds that can be detected.

Growth of sheep from genes polymorphism using artificial intelligence

Polymerase Chain Reaction-Single Strand Conformation Polymorphism (PCR-SSCP) method was used to determine the growth hormone (GH), leptin, calpain, and calpastatin polymorphism in Iranian Balochi male sheep. An artificial neural network (ANN) model was developed to predict average daily gain (ADG) in lambs using input parameters of GH, leptin, calpain, and calpastatin polymorphism, birth weight, and birth type. The results revealed that the ANN-model is an appropriate tool for identifying the patterns of data to predict lamb growth in terms of ADG given specific genes polymorphism, birth weight, and birth type. The platform of PCR-SSCP approach and ANN-based model analyses may be used in molecular marker-assisted selection and breeding programs to design a scheme in improving the efficacy of sheep production.

Sorting apples by watercourse

Before being sent to the market, apples are checked and the ones showing some defects are removed. However, there are also invisible defects that can spoil the apple flavor and look. An example of invisible defect is an internal apple disorder that can affect the longevity of the fruit called a watercore. Apples with slight or mild watercore are sweeter, but apples with moderate to severe degree of watercore cannot be stored for any length of time. Moreover, a few fruits with severe watercore could spoil a whole batch of apples. For this reason, a computational system is under study which takes X-ray photographs of the fruit while they run on conveyor belts, and is also able to analyse (by data mining techniques) the pictures taken and estimate the probability that the fruit contains watercores.

Optimizing pesticide use by data mining

Recent studies by agricultural researchers in Pakistan showed that attempts of cotton crop to yield maximization through pro-pesticide state policies have led to a dangerously high pesticide use. These studies have reported a negative correlation between pesticide use and crop yield in Pakistan. As a result, the excessive use (or abuse) of pesticides is causing the farmers adverse financial, environmental and social impacts. By data mining the cotton, pest scouting data along with the meteorological recordings show how pesticide use can be optimized (reduced). Clustering of data revealed interesting patterns in farming practices along with pesticide use dynamics, helping to identify the reasons for this pesticide abuse.

Explaining pesticide abuse by data mining

To monitor cotton growth, different government departments and agencies in Pakistan have been recording pest scouting, agricultural and metrological data for decades. Coarse estimates of just the cotton pest scouting data recorded stands at around 1.5 million records and growing. The primary agro-met data recorded has never been digitized, integrated or standardized to give a complete picture, and therefore, cannot support decision making. Thus, requiring an Agriculture Data Warehouse. Creating a novel Pilot Agriculture Extension Data Warehouse followed by analysis through querying and data mining, some interesting discoveries were made, such as pesticides sprayed at the wrong time, wrong pesticides used for the right reasons and temporal relationship between pesticide usage and day of the week.

Analyzing chicken performance data by neural network models

A platform of artificial neural network-based models combined with sensitivity analysis and optimization algorithms was successfully used to integrate published data on the responses of broiler chickens to threonine. Analyses of the artificial neural network models for weight gain and feed efficiency from a compiled dataset suggested that the dietary protein concentration was more important than the threonine concentration. The results revealed that a diet containing 18.69% protein and 0.73% threonine may lead to producing optimal weight gain, while the optimal feed efficiency may be achieved with a diet containing 18.71% protein and 0.75% threonine.

Literature

There are a few precision agriculture journals, such as Springer's Precision Agriculture or Elsevier's Computers and Electronics in Agriculture, but those are not exclusively devoted to data mining in agriculture.