Big Data Analytics For Agriculture Input Supply Chain In Ethiopia

Lioutas et al., 2019, Fleming et al., 2018; Kamilaris, Kartakoullis and Prenafeta-Boldú, 2017;Wolfert et al., 2017). As suggested by quite several researchers, a better understanding of themotivations and expectations of the participants is a critical condition for the application of BigData in ASC (Fleming et al., 2018; Jakku et al., 2019; Lioutas et al., 2019)To conclude this section, the research problem that interest this thesis is to understand BDA’spotential in achieving better ASC in Ethiopia. For this to be achieved, it is critical to establish abetter understanding concerning the perspectives of the participants into BDA ASC in the studysetting. As we know it, SCM professionals are the key stakeholders in the supply chain, whoseprimary objective is to effectively manage diverse resources for the realization of sustainablecompetitive advantage. Using a resource-based view, this research intends to understand theexpectations of participants, the requirements and implications of BDA in ASC from theperspectives of participants.1.1.2 BDA in ASC: General IntroductionBDA in agriculture can be understood as a holistic approach to analytics in agriculture, whichcombine weather data, farm-level data, social media data, and market supply and demand data - inorder to generate insight and actionable knowledge to aid data-driven decision making across theentire agriculture supply chain (Fleming et al., 2018; Kamilaris, Kartakoullis and Prenafeta-Boldú2017; Lioutas et al., 2019). As stated by Bronson and Knezevic (2016) the value chain for BigData in agriculture includes from farm input suppliers (i.e. fertilizers, chemicals) to technologyservice providers – each collaborating to drive value from information. Similarly, Ceislik et al.(2018) expressed BDA’s potential in terms of boosting collaboration between the variousstakeholders across the agriculture supply chain. Additionally, Kamble, Gunasekaran, andGawankar (2020) assessed BDA and emerging technologies in agriculture supply chainmentioning their significance in achieving a balance between economic growth, environmentalprotection, and social development – in line with sustainable development goal. In the same vein,Belaud et al. (2019) stretched BDA in agriculture application into the management of farm byproducts - and generating value from agricultural waste in order to improve sustainability alongthe supply chain. Generally, BDA is stated as a revolution that enables the agriculture supply chainto become data-driven and demand-oriented - increasing efficiency, productivity, improving foodsecurity and reducing environmental impacts of agriculture (Lioutas et al., 2019; Kamilaris,Kartakoullis and Prenafeta-Boldú 2017; Fleming et al., 2018; Wolfert et al., 2017).Several studies recently published on BDA in ASC - focusing on agri-food supply chain visibilityand sustainability (Kamble, Gunasekaran, and Gawankar, 2020), carbon footprint and supplierselection (Singh et al., 2018), sustainability and Agriculture by-product management (Belaud et al.,2019) and Agriculture supply chains GIS analytics (Sharma, Kamble and Gunasekaran, 2018) and have provided essential concepts and recommendations on how to improve productivity andefficiency. Beyond productivity and efficiency, more recent articles focused on sustainabilitybenefits of BDA in ASC (Allaoui et al., 2018; Kamble, Gunasekaran and Gawankar, 2020).As revealed by quite several existing research on BDA in the supply chain, it is critical tounderstand how BDA interplay with the various supply chain variables in the context, as to helpdrive real value from the service. Understanding the variables requires addressing environmental,social, technological and economic factors in the context (Kamble, Gunasekaran and Gawankar,2020). Notably, it is critical to understand how to achieve coordination among the members of thesupply chain network (Papadopoulos et al., 2017, Kamble, Gunasekaran and Gawankar, 2020),how to use BDA for Supply chain sustainability initiatives (Hazen et al., 2016), how to improve6

supply chain performance (Gunasekaran, 2017), how unstructured data can be used and createvalue in supply chain processes and network (Chen, Preston and Swink, 2015; Papadopoulos etal., 2017) how to achieve traceability in the stream -up and down- of the supply chain (Zhu et al.,2018) and what technology, infrastructure and resources required to drive the supply chain network(Zhong et al., 2016), issues of data quality (Hazen et al., 2014) and the knowledge and skillsrequired for data analytics (Wang et al., 2016, Waller and Fawcett., 2013).1.1.3 BDA in ASC: Studies in EthiopiaIt is widely believed that the adoption of BDA will enable developing countries AISC to be moreefficient, productive and sustainable (Kamble, Gunasekaran and Gawankar, 2020; Kamilaris,Kartakoullis and Prenafeta-Boldú, 2017; Wolfert et al., 2017). However, for this to be achievedthe motivation and expectations of stakeholders in the context need to be understood (Kamble,Gunasekaran and Gawankar, 2020; Lioutas et al., 2019; Fleming et al. ,2018; Jakkua et al. 2019).However, BDA in AISC is a new phenomenon, and the available studies on the area addressedmostly developed countries (Kamble, Gunasekaran and Gawankar, 2020; Kamilaris, Kartakoullisand Prenafeta-Boldú, 2017; Wolfert et al., 2017), and from the literature, we reviewed so far, nostudies conducted on BDA In AISC in Ethiopia. Nevertheless, we found three articles whichmention - Big Data Analytics, Agriculture, and Ethiopia. The first two (McCarty et al., 2017; Neighet al., 2018) dealt with satellite area mapping study using Big Data - which is not in our scope. Thethird one (Akal et al., 2019) is a Big Data case study on four organizations in Ethiopia - which isnot in our scope again. Moreover, it worth noting that though Akal et al. (2019) seems to set outto research BDA, their research mostly dealt with structured data, and the paper has no mention ofunstructured data such as social media data. However, Akal et al. (2019) have listed severalchallenges for the application of BDA in Ethiopia including data quality problem, lack of datacollection and handling standards, knowledge and awareness issues and lack of management focus.To conclude this section, for BDA in AISC to achieve its intended goal, it is critical to understandthe topic from the participants perspective, and the topic in Ethiopia context remained unexplored.A broader understanding of the expectations of participants, the requirement and implications ofBDA in AISC might shed light on practical challenges of the industry, and help to understandpractice and knowledge gaps more clearly.1.2 Purpose Statement and Research Questions1.2.1 Purpose StatementThe purpose of this qualitative study is to explore the potential application of BDA in AISC fromthe perspectives of SCM professionals. The study aims to understand the expectations ofparticipants, the requirement and implications of BDA in AISC from the participants perspective.At this stage in the research process BDA for AISC in Ethiopia defined as the application ofvarious data analytics techniques in to structured and unstructured data - in order to generateinsight and actionable knowledge to aid data-driven decision making for supply of input inagriculture.1.2.2 Research QuestionsBased on the purpose of the study, this research asks the following questions from the perspectivesof the SCM professionals in the context:7

RQ1 - What are the expectations of the participants in the application of BDA in AISC?RQ2 - What are the requirements for applying BDA in AISC?RQ3 - What are the potential implications of applying BDA in AISC?1.3 Topic JustificationThis research might interest several public and private organizations. Considering the potentialimplications of BDA in the agriculture supply chain, the study might provide useful insight to dealwith the current food security and sustainability challenges in the country. A better understandingof the expectations of the participants might provide useful insight to motivate the adoption of thetechnology. It may also generate useful inputs for policy development. In addition to societalinterest, the research may be used by technology and agribusiness companies to understand theparticipant’s requirement - and this understanding may provide useful insight on how to developBDA enabled solutions tailored to the context.Considering the unavailability of prior research in the context (see section 1.3) - the study mightprovide valuable contribution to the scholarly research and literature in the field of study.1.4 Scope and LimitationsThe participants for this study were purposefully selected supply chain management professionalswhose work is related to agricultural input supply in Ethiopia. The participants work for fertilizer,seed, crop protection and agribusiness organizations in the country. The study focused on theperspectives of the participants in the study topic. The study addressed the challenges andpotentials in the topic area; however, the research did not address how to apply the technology inthe research setting. The documentary examination task only focused on relevant materialspublished since 2015 - to help the researchers address the recent trends in the domain.This research is subject to a number of limitations. BDA is a new phenomenon, particularly inEthiopia. Thus, the unavailability of prior research in the country is one of the limitations of thisresearch. The research focused on a new subject with an unknown element, and we chose to followa qualitative exploratory approach, and the data was collected through semi-structured interviewsand document analysis. Based on this the research has taken limitations which comes with theselected data collection methods - for example, the presence of the researchers in the interviewprocess might have created biases on the responses of the participants (Creswell and Creswell,2018).The short time frame for the project was undoubtedly one of the limitations of this thesis; Thus,the findings of the study is limited to the data which was collected and analyzed in the limited timeframe.1.5 Thesis OrganizationThis thesis has six sections. And the organization of the thesis explained as follows:Section 1 – This is the introduction part - Here we introduce the background, the researchsetting, the topic area, purpose statement, the topic justification, the scope, limitations andthe organization of the thesis.8

Section 2 – In the Literature review part, we discuss the literature search process, themethod applied, the relevant concepts and theories from past studies in the topic. Moreover,our theoretical approach will be discussed in this part.Section 3 – In the methodology section, we discuss our approach and the research processin detail, including ethical considerations.Section 4 – In the empirical finding section, we present the data and findings from theinterview, documentary analysis.Section 5 – In the discussion section, we present the analysis with the findings of primarydata, secondary data and literature review.Section 6 – In conclusion, the contribution of the study to the theory, practical applicationincluding impact on industry, policy and recommendation on further research will bepresented.9

2Review of the LiteratureIn this section, we first explain how we conducted the literature review. Then we present the basicconcepts found from the literature review. Finally, the theoretical framework for the study will bepresented.Abbreviations: RBV – Resource Base View, ASC – Agriculture Supply Chain, AISC- AgricultureInput Supply Chain, SCM – Supply Chain Management, GIS – Geography Information System,GPS - Global Positioning System, OLAP- Online analytical processing2.1 Literature search ProcessIn this thesis, we follow Levy and Ellis (2006) systematic framework to conduct an effectiveliterature review. The objective is to analyze and synthesize high-quality peer-reviewed articlesand establish a strong foundation for the proposed topic and methodology; and, to justify that theproposed study provides a novel contribution to scholarly research and literature in the field ofstudy (Levy and Ellis, 2006). Based on this the literature review is conducted in three stages: Inputs- literature collecting and screening, processing based on bloom's taxonomy and output - writingthe final literature review - output (Levy and Ellis, 2006).2.1.1 Inputs - Literature collection and screening processIn the literature collection and screening stage, a combination of 13 different keyword searcheswere carried out in Scopus database. The search is limited to peer-reviewed articles published inthe last 10 years. The year limitation is based on the fact that the research topic is a phenomenonin the last 10 years. Based on the search, a total of 61 articles were found, of which 5 articlesrepeated several times (see appendix 1). With the inputs process going, 8 more articles are foundwhich contained much richer aspects from both horizontally and vertically in the researcher area.For example, the BDA applications in AISC, different type of analysis which all will bediscussed in the following sections.2.1.2 Processing - Based on bloom's taxonomyIn this part the literature process was conducted in accordance with Bloom's taxonomy whichinclude knowing and comprehending the literature, applying, analyzing, synthesizing andevaluating (Levy & Ellis, 2006). The selected articles are initially filtered by the relatedkeywords, then after read the abstracts and main frame of the articles, more articles filtered. Inthe middle of the filter process, more articles added in the list which are picked from thereference list of corresponding articles. In order to conclude as much as whole picture of ourresearch area as well as make the gap above the sea.2.1.3 Output - Writing the final literature reviewThe literature review mainly conducted from answering a few questions, such as: What are thebasic concepts of our research area? How do these concepts apply under different background?What’s the connections between them? What are the challenges and gaps of BDA in AISC? Andhow RBV connects with BDA and AISC? Then some barriers and challenges are proposedregarding the research questions.10

2.2 Basic Concepts2.2.1 Big DataModern digital technologies can better understand complex agricultural ecosystems and meet theincreasing challenges of agricultural production. These technologies can continuously monitor thephysical environment and generate large amounts of data at an unprecedented rate (Kamilaris,Kartakoullis and Prenafeta-Boldú, 2017). These technologies generate large amounts of data,called big data, for example, there is data from continuous measurement and monitoring of thephysical environment, sensors on fields and crops provide granular data points on soil conditions,and detailed information on wind, fertilizer requirements, water availability and pests (Nidhi,2020). These all belong to the scope of smart agriculture, smart agriculture helps to automateagriculture, collect data from the field, and then analyze it, this can help farmers decide to plantthe right crop at the right time and achieve the purpose of making an informed decision (Nidhi,2020). Big data in agriculture also has the dimensions of Volume, Velocity, Variety, Veracity,Value, but big data is notorious for its accuracy and stability, so from such a large amount of data,how to extracting information and doing the accurate prediction in a reasonable time is the key,this behaviour is also known as big data analysis (Nidhi, 2020 ; Kamilaris, Kartakoullis andPrenafeta-Boldú, 2017).2.2.2 Big Data AnalyticsData volume is not the biggest issue in big data, the sources of data are mostly heterogeneous, thevolume and speed of data are also different (Kamilaris, Kartakoullis and Prenafeta-Boldú, 2017).They are expressed in different types and formats, and the access to data is also different (forexample, web services, Repositories, feeds, files, archives, etc.), these facts all point the issue inone direction: the ability to search, aggregate, visualize and cross-reference large data sets in areasonable time. It is about the ability to extract information and insights, that is, the big dataanalysis which also mentioned above(Kamilaris, Kartakoullis and Prenafeta-Boldú, 2017). Whenbig data analytics is linked to agricultural SCM, new challenges arise.2.2.3 Big Data in Supply Chain ManagementBig data gradually becomes an important information technology regarding agricultural foodsupply chain decisions (Ahearn, Armbruster and Young, 2016). Big data cannot be used separatelyfrom society and ethical issues and also is actively constructed and explained by people, especiallyin social and technological environments (Bronson and Knezevic, 2016). Especially today's supplychain professionals are overwhelmed by massive amounts of data, which has inspired people tothink about new ways to generate, organize, and analyze data. This provides an impetus fororganizations to adopt and improve data analytics functions such as data science, predictiveanalytics, and big data to enhance supply chain processes and ultimately improve performance(Hazen et al., 2014). This is a very complex workflow that needs to be integrated across differentdisciplines. For example, information systems experts to gain insight into how to collect, store,process and retrieve data. SCM experts need to ensure that the analysis being performed is thecorrect problem, and the results of the analysis are relevant, etc. (Hazen et al., 2014). This paperis based on this purpose, hoping to link BDA and SCM in the Agri-food area.11

2.2.4 Big Data Analytics in Agriculture Supply ChainFirst of all, it is certain that the analysis of these (big) data will enable farmers and enterprises toextract value from it, thereby increasing their productivity (Kamilaris, Kartakoullis and PrenafetaBoldú, 2017). The scope of big data is not limited to agricultural production but also affects theentire food supply chain, big data needs to be unlocked smartly, analysis has the potential to addvalue at every step, it can start by choosing the right agricultural inputs, monitoring soil moisture,tracking market prices, controlling irrigation, finding the right point of sale, and getting the rightprice to processing value chain (Nidhi, 2020). Coincidentally, Wolfert et al., (2017) also mentionedthat the application of big data in smart agriculture goes beyond primary production, it is affectingthe entire food supply chain, and it is being used in many aspects, for example, providing predictiveinsights into agricultural operations, driving real-time operational decisions, and redesigningbusiness processes for game-changing business models.2.2.4.1 Social, environmental and economic aspectsKamble, Gunasekaran and Gawankar (2020) mentioned that emerging technologies such as BDAare pushing the traditional AISC towards a data-driven digital supply chain environment. In thistransformation process, we must consider not only food production methods, but also socialconcerns, environmental concerns, food safety and quality requirements, and economic feasibility(Kamble, Gunasekaran and Gawankar, 2020). Also, these solutions should not be limited toagricultural production but should cover the entire supply chain, including food processing,packaging, distribution and consumption (Kamble, Gunasekaran and Gawankar, 2020). This willallow decision-makers to have the information they need to develop a sustainable supply chainstrategy (Kamble, Gunasekaran and Gawankar, 2020). When focusing on the environment, inaddition to the control of the natural environment, the political environment was also mentioned.Issues such as insufficient land use, high rents, limited processing capacity, and hostile politicalenvironments are major challenges in the development of alternative supply chains (Ka

2.2.3 Big Data in Supply Chain Management 11 2.2.4 Big Data Analytics in Agriculture Supply Chain 12 2.2.4.1 Social, environmental and economic aspects 12 2.2.4.2 Big data Analytics applications in the Agriculture Supply Chain Process 13 2.2.4.3 Analysis in Supply Chain Management: 14 2.2.4.3.1 Descriptive analytics: 15