Literature Survey

People nowadays are so preoccupied with their daily tasks that they have very little time to complete all of their duties. As a result, individuals are more likely to use items that assist them save time when doing daily chores. They have become increasingly reliant on machinery rather than performing things by hand. This approach has recently gained popularity in the agriculture industry as well. Farmers had to utilize a lot of human strength in the past to meet their agriculture demands. However, utilizing a single machine and only one person to manage it, the same amount of work may now be accomplished in less time and for less money. Farmers are unable to locate appropriate markets and receive enough fluctuation details on demand and prices for their cultivations. Furthermore, they are less knowledgeable about the optimal growing conditions for crops and the best crops for current areas. As we can see from the background part, there are many systems in place for agricultural-related operations. However, IT-based technologies are rarely employed in Sri Lanka. The majority of the systems are imported from other nations. As a result, it is critical to continue performing research and development in this sector. To alleviate these issues. It's crucial to understand existing applications before coming up with a solution. As shown below, we compared several research articles to our suggested system.

01. Forecasting price for specific vegetables and forum for knowledge sharing.

Research on hybrid neural network and H-P filter model for short-term vegetable price forecasting was carried out and the focus of this research is on time series data on vegetable costs, which have a significant impact on people's lives. In people's daily life, an accurate price forecasting approach and an early-warning system in the vegetable market are critical. There are both linear and nonlinear patterns in the time series pricing data. As a result, neither a contemporary linear forecasting model nor a neural network can be used to model and predict time series data. The linear forecasting model cannot handle nonlinear relationships, while the neural network model cannot manage both linear and nonlinear patterns simultaneously. From time series data, the linear Hodrick-Prescott (H-P) filter may extract the trend and cyclical components. They have forecasted both linear and nonlinear patterns, then linearly mix the two components to give a forecast from the original data. The topology of a hybrid neural network based on an H-P filter that learns the trend and seasonal patterns separately is proposed in this study. The model is evaluated using data from vegetable prices in the trial. According to this research paper their method outperforms the autoregressive integrated moving average method and back propagation artificial neural network methods, according to comparisons [4]. A study on prediction of vegetable price based on neural network and genetic algorithm was conducted and they have used four distinct types of models. • BP neural network model • The neural network model based on genetic algorithm • RBF neural network model • An integrated prediction model based on the three models above The four models are used to forecast the price of Lentinus edodes for the Xinfadi wholesale market in Beijing. 84 total records have used to train and test the four models, which were collected between 2004 and 2009. In conclusion, the BP neural network model has the weakest predictive ability. The RBF neural network model was generally more accurate than the genetic algorithm neural network model. The outputs of the integrated prediction model are the best. [5] Another research was done on Alternative Forecasting Techniques for Vegetable Prices in Senegal. Based on this research paper Dr Alioune DIENG evaluate the effectiveness of parametric models for projecting specific vegetable prices and make recommendations to potential users. There are two approaches of predicting that are used. The methodologies' forecasts were evaluated using both qualitative and quantitative criteria. Three alternative parametric models and a non-parametric model are considered as forecasting methodologies. The naïve, exponential, and Box and Jenkins autoregressive integrated moving average (ARIMA) models are among the parametric models. The spectrum analysis technique is used in the non-parametric model. According to the findings of this study, among parametric models, Box and Jenkins' autoregressive integrate moving average model will be a good technique to utilize in providing vegetable price forecasts for producers and consumers. But more study is needed to compare the accuracy of parametric and non-parametric models in forecasting other crops [6]

02. Forecasting demand for specific vegetables and Dashboard for users.

It is critical to understand existing applications before developing a solution. Based on performance analysis, a study was conducted to select the appropriate forecasting model at the retail stage for selected vegetables.[7] Various forecasting models, including the Box–Jenkins-based autoregressive integrated moving average model, and machine learning-based algorithms, including long short-term memory (LSTM) networks, support vector regression (SVR), random forest regression, gradient boosting regression (GBR), and extreme GBR (XGBoost/XGBR), were proposed and applied at the retail stage for selected vegetables to forecast demography. The performance analysis was carried out to select the best forecasting model for selected vegetables at the retail stage. Based on the results obtained for a case environment, it was discovered that the machine learning algorithms, namely LSTM and SVR, produced superior results when compared to other different demand forecasting models. Implementing LSTM and SVR for the case scenario at the retail stage will reduce forecast error, daily retail inventory, and fresh produce waste, while increasing daily revenue.[7] Another study focused on managing change and addressing the different expectations of domestic and foreign stakeholders in the context of Indian agriculture.[8] Successful forecasting can be a useful tool to accomplish the above-mentioned aims, especially when dealing with vegetables with a limited shelf life. Based on previous data, the researcher forecasted the daily demand for the fresh vegetable product (onions) in a Mumbai wholesale market during the study. A seasonal autoregressive integrated moving average (SARIMA) model surpassed other contenders in terms of forecasting accuracy on both in-sample and two out-of-sample datasets among the models designed and tested. The model's results demonstrate that it can forecast with a mean absolute percentage error (MAPE) of 14%, which is regarded acceptable for products with stochastic demand like fresh vegetables.[8] A recent research paper on China's agriculture output values forecasting a novel two-stage model that combines the grey first-order differential equation model with genetic programming was developed to anticipate the value of agricultural imports based on the Grey Seasonal Model.[9] For agricultural management prediction, a completely probabilistic strategy using dynamical ocean-atmosphere models was used to anticipate climatic variability at seasonal and interannual time scales. For agricultural production forecasting and enhanced grey forecasting model based on a genetic algorithm was developed. Particle swarm optimization paired with the grey linear power index model, the grey logarithm power model, and the grey parabola power model was used to predict grain production in China. To anticipate the agricultural areas in China affected by natural catastrophes, a nonlinear autoregressive exogenous model with a neural network and grey system model was combined. For predicting data having seasonal characteristics, there are two basic groups of algorithms now in use. Researchers in this work used grey cumulative generation to reduce data fluctuations before developing a grey seasonal model (GSM). By comparing processes, SARIMA and the Holt-Winters model surpass each other.[9]

03. Identify the best cultivation for existing land.

A Fuzzy Based Decision Support System for Evaluating Land Suitability and Selecting Crops was carried out and the focus of this research is giving the best suggestion for farming in lands using few variables. Such as climate, landscape and soil, topography, wetness. The proposed system has six phases. a) identification of the decision maker's requirements. b) determination of membership functions of each criterion. c) determination of performance rating of each land limitation. d) determination of performance rating of landscape and soil hmita1ion. e) evaluation of land suitability class. f) selecting the appropriate crop. All land characteristics are considered as having defined fuzzy sets. Hence the real-valued input variables arc transformed into fuzzy sets. This stem is applied to each land characteristic factor considered m the solution of the problem. The next step is the inference process it relates systematically pairwise all the factors that take place in the solution depending on the purpose of the problem. This part includes many fuzzy conditional statements to describe a certain situation. Land suitability class 1s determined through a two-phase inference based on input data expressed as crisp value and fuzzy set. First, the inference process is done to set the limitation level and secondly, it is done to determine the suitability class of the land. [10] Another study on Geographic information system-based identification of suitable cultivation sites for wood-cultivated ginseng was done. Wood-cultivated ginsengs are perennial plants that are semi-heterophonic and commonly used in Chinese medicine. To identify suitable sites for the propagation of wood-cultivated ginseng, factor combination technique (FCT) and linear combination technique (LCT) were used with a geographic information system and the results were superimposed onto an actual wood-cultivated ginseng plantation. The LCT more extensively searched for suitable sites of cultivation than that by the FCT; further, the LCT probed wide areas considering the predominance of precipitous mountains in Korea. In addition, the LCT showed a much higher degree of overlap with the actual cultivation sites; therefore, the LCT more comprehensively reflects the cultivator’s intention for site selection. On the other hand, the inclusion of additional factors for the selection of suitable cultivation sites and experts’ opinions may enhance the effectiveness and accuracy of the LCT for site application. [11] Integration of artificial neural network and geographical information system for an intelligent assessment of land suitability for the cultivation of a selected crop is another study which was carried out in this domain. The main objective of this study is to investigate the potential of artificial neural networks (ANN) for integration with geographical information systems (GIS) to assess the suitability of the land to cultivate a selected crop. For this purpose, requirements of a system for an intelligent assessment of land suitably were determined and the architecture of the integrated system was designed according to the capabilities of ANN and GIS. [12]

04. Favorable conditions based on the crop.

A study on Best Crop Rotation Selection with GIS-AHP Technique Using Soil Nutrient Variability was done earlier. Crop selections and rotations are very important in optimizing land and labor productivities, enhancing higher cropping intensities, producing better crop yield. This is a land suitability analysis system based on the analytical hierarchy process (AHP) technique coupled with the Geographic Information System (GIS) software environment that can be a unique tool for better crop selection. The AHP-GIS technique was used in land suitability analysis in crop rotation decisions, for rice-jute (Kharif season) and potato-lentil (Rabi season) crops in the Hooghly District, West Bengal, India. The study area covering 291 ha was classified based on eight major soil nutrient levels with 70 randomly selected plots for soil sampling and analysis. The soil nutrient variability was examined with descriptive statistics followed by the best semi-variogram-based model selection for kriging interpolation in the ‘R’ software environment. The pairwise comparison matrix based on the ranking of parameters and giving weights were carried out considering the importance of each parameter for specific crops. [13] Another study on Modelling crop diversification and association effects in agricultural systems was carried out in this domain. The need to redesign more sustainable agricultural systems able of producing more, especially through intercropping or agroforestry, cannot be achieved without taking into account the essential aspect of production variability. In here a shift towards sustainable intensification of agricultural systems, considers the dual dimensions of yield and risk in a combined framework for the assessment and the comparison of two diversification strategies: (i) a simple diversification strategy (SDS) considered as an increasing number of crops grown on separate plots within a farm and (ii) an intercropping strategy (IC) considered as a within-plot increased diversity, where more than one species is grown at the same time and place. The two perspectives examined here were Modern Portfolio Theory and Land Equivalent Ratio. The former quantifies the effect of diversification on risk, the latter measures the effect of association on production. This research merges both approaches in a combined framework to assess intercropping system performances. By applying our framework to cases selected from the literature, we explored and compared the potential benefits of these two strategies in terms of yield and risk. Results showed that intercropping, in addition to being interesting concerning yield, can have an additional risk reduction effect compared to a simple diversification strategy. Conversely, some crop mixtures maintained or even increased yield variability. This work contributes to a better understanding of the possible impacts of diversification strategies on trade-offs between yield and risk but also underlines the importance of taking yield variability into account in further studies. [14] A Fuzzy Based Decision Support System for Evaluating Land Suitability and Selecting Crops is an evaluating land suitability and selecting crops in modem agriculture is of critical importance to every organization. This is because the narrower area of land, the more effective in planting 1s required following the desires of the land. The process of evaluating land suitability class and selecting plants by decision marker's requirements is complex and unstructured. Approach: This study presented a fuzzy-based Decision Support System (DSS) for evaluating land suitability and selecting crops to be planted. A fuzzy rule was developed for evaluating land suitability and selecting the appropriate crops to be planted considering the decision maker's requirements in crops selection with the efficient use of the powerful reasoning and explanation capabilities of DSS. The idea of letting the problem be solved determines the method to be used was incorporated into the DSS development. As a result, effective decisions can be made for land suitability evaluation and crop selecting problems an example was presented to demonstrate the applicability of the proposed DSS for solving the problem of evaluating land suitability and selecting crops in real-world situations. [15]

Research Problem and Gap

Although Sri Lanka is a fertile tropical land with great potential for the cultivation of a wide range of crops, the agriculture sector is hampered by deep-rooted productivity and profitability problems. Mechanization adaptation is low, as is a private agricultural investment. Natural disasters such as floods, droughts, and animal attacks confront the farming community every year, wreaking havoc on their lives and livelihoods, resulting in a high suicide rate among farmers. Even though Sri Lanka is geographically, intellectually, and biologically well-positioned to conduct research and development, little is being done in this field. When we looked at the primary issues in Sri Lankan agriculture, we observed a common set of predicaments, which are listed below categorizing to respective components,

01. Forecasting price for specific vegetables and forum for knowledge sharing.

Although Sri Lanka is a fertile tropical land with great potential for the cultivation of a wide range of crops, the agriculture sector is hampered by deep-rooted productivity and profitability problems. Mechanization adaptation is low, as is a private agricultural investment. Natural disasters such as floods, droughts, and animal attacks confront the farming community every year, wreaking havoc on their lives and livelihoods, resulting in a high suicide rate among farmers. Even though Sri Lanka is geographically, intellectually, and biologically well-positioned to conduct research and development, little is being done in this field. When we looked at the primary issues in Sri Lankan agriculture, we observed a common set of predicaments, which are listed below, • Post-harvest wastage since could not find retailers/consumers to sell organic products as farmers are not cultivating to meet demands. • Insufficient importance on the development and production of nutrient-dense foods. • Food preferences are shifting, as are concerns about how it is grown, with a greater focus on ethically sourced organic crops. • Lack of traceability of products. • Farmers abandon their plantations due to their incapacity to combat the various pests that attack them. • Lack of market linkages and inability to analyze pricing due to lack of knowledge of market ups and downs. When it becomes to Vegetable Price Prediction in Sri Lanka there is not any other source to get the market price forecasting for farmers and consumers. Due to this situation farmers face several issues. • Could not sell the product getting a good profit. • Sometimes price goes to lower than the cost. • Products will become unconsumable when they store them expecting market price fill raise.

02. Forecasting demand for specific vegetables and Dashboard for users.

The distribution of fresh vegetables has been a major challenge in recent years. The biggest cause of vegetable waste, according to farmers, is their failure to estimate demand for veggies. Due to the absence of forecasting of vegetable demand, both growers and retailers are inconvenienced. When examining the retailer's perspective, the market might be overstocked and understocked at times due to unanticipated demand. As a result, a projected model is required to reduce vegetable waste. We must consider the weather in this case. Because the quantity demanded may fluctuate depending on the weather conditions. When it comes to weather, regional differences are important because different parts of the country have different sorts of weather. As a result, there are some differences in how vegetables are grown in different parts of the country. Seasonal influences and weather should be understood because they have a significant impact on demand forecasting. Import foods, promotions, and business vegetable production all have an impact. As a result, the amount of demand for a certain vegetable may change. Another factor to consider in this dilemma is local demand. Farmers in supermarkets should be informed about the required quantity of specific production at least two days ahead of time because they normally forecast daily. So, to order the exact number of the particular veggie item, we'll need a means to accurately forecast the required quantity. So that we can cut down on veggie waste. The availability and price have an impact as well. They are important considerations in forecasting demand. It's critical to think about the amount of food that farmers can provide. It can be viewed as yet another effect of demand forecasting.

03. Identify the best cultivation for existing land.

Even though Sri Lanka is a lush tropical region with a lot of potential for growing a broad variety of crops, the agricultural industry is limited by low productivity. Even though modern agriculture is partly intermingled with technology but could not see few advances in information technology with agriculture in Sri Lanka. This might be due to farmers' inability to raise their production capacity and quality of the product to meet market demand, • Their failure to effectively manage their land. • Their inability to cultivate without first knowing the land's character. • Use of various chemical fertilizers available in the market. • Low demand for organic fertilizer products. • Intense competition in the market.

04. Favorable conditions based on the crop.

Although Sri Lanka is a fertile tropical land with great potential for the cultivation of a wide range of crops, the agriculture sector is hampered by deep-rooted productivity and profitability problems. Mechanization adaptation is low, as is a private agricultural investment. Natural disasters such as floods, droughts, and animal attacks confront the farming community every year, wreaking havoc on their lives and livelihoods, resulting in a high suicide rate among farmers. Even though Sri Lanka is geographically, intellectually, and biologically well-positioned to conduct research and development, little is being done in this field. When we looked at the primary issues in Sri Lankan agriculture, we observed a common set of predicaments, which are listed below, • Post-harvest wastage since could not find retailers/consumers to sell organic products as farmers are not cultivating to meet demands. • Insufficient importance on the development and production of nutrient-dense foods. • Food preferences are shifting, as are concerned about how it is grown, with a greater focus on ethically sourced organic crops. • Lack of traceability of products. • Farmers abandon their plantations due to their incapacity to combat the various pests that attack them. • Lack of market linkages and inability to analyze pricing due to lack of knowledge of market ups and downs. What the farmers do is they decide the crops by thinking about their parents’ cultivation methods and continue their opinions. But sometimes this does not work. With the climate and geographical change, the requirements will differ from time to time. So it is a serious conflict to decide the best crop to cultivate. This project will guide them to decide the best crop in a specific period.