The purpose of this study is to predict the factors affecting the readiness to adopt big data in small and medium enterprises (SMEs) in Vietnam. Data collected from 240 managers at SMEs encompasses 13 input variables that impact the readiness (high/low) to adopt big data. Partitioned, training, and testing data were analyzed by three Data Mining algorithms techniques (CHAID, Bayesian networks, and Neural Network). The accuracy results of evaluation statistics on the training and testing data of the three models are 70% higher. The area under the Receiver Operating Characteristics (ROC) curve (AUC) value on the training data ranged from 0.827 to 0.908, while it ranged from 0.777 to 0.898 on the testing data. The results of this study highlighted that top management support, data quality, data security, partner pressure, and budget resources are the five most important factors to predict readiness. The findings of this study contribute important implications for managers, service vendors, and policymakers to understand the factors that influence readiness to adopt big data. Hence, managers can establish a clear strategy to enhance the readiness to adopt big data in SMEs in the future.
The rapid development of internet technology has created a huge amount of data from many different sources such as Media, Cloud, Web, Internet of Things, and Databases that are called big data. Big data is a "5V" characterized data source, exhibiting volume, velocity, variety, verification, and value characteristics [1]. Big data adoption is useful in helping companies seek new business opportunities, decrease costs, and minimize risks [2,3]. The use of big data in e-commerce context provides many benefits such as prediction trends for future product development and improvement of company-customer relationships [4]. Further, big data plays an important role in helping businesses develop a sustainable economy [5]. However, firms are facing many challenges when implementing big data including, information technology infrastructure, financial resources, data security, organizational culture, lack of skills, etc. [2,6-8]. In Vietnam, SMEs account for 98% of total enterprises, 30% of total export value, and create 500,000 new jobs annually [9,10]. Moreover, Vietnam is considered to be the place where large data sources are available with 66% and 60% of the population using the internet and using social networks, respectively [11]. However, SMEs in Vietnam have not yet taken full advantage of new information technology such as cloud computing, internet of things, and big data, thus faced with low business efficiency [12]. The big question is, what are the factors that affect the adoption of big data by SMEs in Vietnam? Leveraging this appeal, this research sought to predict the factors that impact the adoption of big data by SMEs in Vietnam. Understanding the factors that affect the readiness to adopt big data helps businesses to be well prepared before adopting big data. As a result, the likelihood of successful big data adoption will be high. Therefore, the adoption of big data will contribute to improving enterprises’ business performance. 2
Previous studies have mentioned several factors influencing big data adoption (BDA).
Yadegaridehkordi, et al. [13] applied the approach of decision-making trial and
evaluation laboratory (DEMATEL) and adaptive neuro-fuzzy inference systems (ANFIS) to
show that big data adoption in Malaysian manufacturing firms was mostly affected by
factors such as perceived advantage, complexity, technology resources, big data
quality, and integration. Verma and Bhattacharyya [
BDA in South Africa’s public service sector and discovered that technological infrastructure, security, reliability, finances, competitors, customers, and vendor support factors are prerequisites for assessing the readiness to analyze big data. Klievink, et al. [18] conducted a study to evaluate the readiness of businesses in the Dutch service sector as affected by organizational alignment, maturity, and capabilities. Mneney and Van Belle. [19] evaluated four categories (technology, organization, environment, and task technology fit) that influence retail organizations in South African to apply big data.
Based on the literature review, it can be seen that there are many related studies on the factors affecting the adoption of big data. However, the factors influencing the big data adoption readiness in SME manufacturing and service sectors in South East Asia have not been evaluated. Moreover, the application of Data Mining technique to predict factors affecting readiness to adopt big data is still rare. 3 3.1
Questionnaires were divided into two parts. In the first part, 35 items were included to evaluate factors that impact the readiness to apply big data in SMEs, and 9 items to assess the readiness to apply big data. In the second part, the questionnaires inquired about the firms' information and respondents’ socio-economic characteristics. The seven-point Likert scale used ranged from 1 for "strongly disagree" to 7 for "strongly agree". The questionnaire's reliability and validity was assessed through a pilot test. The 30 respondents were composed of big data experts, Chief Executive Officers of companies, and professors. Cronbach's alpha was used to evaluate the internal consistency within the data. The pre-test analysis revealed Cronbach alpha values greater than 0.7 for all questionnaire items. Subsequently, the questionnaire was slightly modified to fit the reality of the company. Responses from this pilot study were not included in the final sample.
The subjects of this study are SMEs which are distributed over six areas of Vietnam’s main sub-industries: food and beverages, construction, garment, wholesale, retail, accommodation services survey. The questionnaires were sent by emails to qualified individuals in SMEs in Vietnam. Finally, 240 managers were chosen from manufacturing and service companies, and data were collected at the end of 2020. According to the descriptive statistics, the majority (46.7%) held Bachelor’s degrees and 39.2% had Master's degrees or higher. The respondents were 72.5% male and 27.5% female, with more than half (57.9%) between 31 to 45 years of age, and belonging to either small enterprises (82.5%), medium enterprises (17.5%). 3.2
Cronbach's α value was calculated to evaluate the reliability of the scale. The analysis results showed that the Cronbach's α value of all variables ranged from 0.626 to 0.867 (Table 1). Hair, et al. [20] suggested that Cronbach’s α value should be higher than 0.7, but 0.6 is acceptable. Therefore, the results prove that all variables in the study are consistent and reliable enough for further analysis.
nuImtebmers Cronbach α New measurement
To predict the factors’ influence on readiness, the output variable (readiness to adopt big data) was categorized into two levels: high and low. Based on an average of 9 items used to identify the readiness to adopt big data in SMEs, the output variable was code conducted. "High" with the mean value of 9 items higher than or equal to 6.0 and "low" with the mean value of 9 items lower than 6.0. The thirteen input variables include: relative advantage, top management support, organization culture, technical competence, budget resources, data quality, data security, competitive pressure, IT infrastructure, partner pressure, government support, firm size, and type of industry. To classify the readiness of SMEs to adopt big data, this study used three techniques of data mining including CHAID, Bayesian networks, and Neural Network with SPSS modeler 18 software. These algorithms are common algorithms in Data mining techniques.
Giang et al.
In Table 2, the detection sensitivities of CHAID, Bayesian networks, and Neural network classifiers for training data were 83.54%, 79.88%, and 74.39%, respectively. The results of the prediction on the testing data were 71.05%, 81.58%, and 72.37%, respectively. The accuracy of all 3 classification models is higher than 70.00%. According to Tavakoli [28] these models have high prediction accuracy. This suggests that 13 input variables including relative advantage, top management support, organization culture, budget resources, technical competence, data quality, data security, competitive pressure, IT infrastructure, partner pressure, government support, type of industry, and firm size were well identified factors that impact Vietnamese SMEs’ readiness to adopt big data.
Area under the ROC curve (AUC) are used as appropriate measures to evaluate the performance of classification algorithms and have values ranging from 0.5 to 1. The model will be considered to have acceptable discrimination if the AUC value is higher than 0.7 [28]. In this study, the AUC value on the training data ranges from 0.827 to 0.908 and on the testing data range from 0.777 to 0.898 meaning that the models would be considered to have good discrimination [28]. The AUC for training data of CHAID, Bayesian networks, and Neural Network algorithms were 0.907, 0.908, and 0.827, respectively. While the respective figures for testing data were 0.777, 0.898, and 0.780. The results show that the AUC value was highest with Bayesian networks on both training data and testing data models.
Today's adoption of big data offers many benefits to companies including increased performance, improved strategic direction, development of more reliable customer service, product identification and development, and reducing risks [2,3,29]. The aim of Top management
support Data security
Government
support IT infrastructure 1 2 3 4 5 6
support Data quality Organizational culture Partner pressure
Firm size 1 2 3 4 5 6 7 8 9 10 the study is to predict the factors that significantly influence the readiness of SMEs in Vietnam to adopt big data. Research results show that the five factors that have the most important influence on the readiness to apply big data include top management support, data quality, data security, partner pressure, and budget resource. Top management support plays a key role in the adoption of big data as it guides resource allocation, integration of services, and redesign of processes [29]. Top management has clear goals for adopting big data, and encouraging building a big data-driven decision-making culture that will play an important role in improving the quality of the company's decisionmaking [27,30]. Therefore, if top management understands the benefits of big data adoption, they will invest resources and encourage employees to implement big data which is consistent with previous researches [15,29]. Data quality directly affects the results of the business analyses [31]. Big data come from different sources such as databases, text data, graph, documents, images, videos, audio files, emails, comments, tags, tweets, and clicks, etc. [6,32]. SMEs are abundant in data sources and high accuracy which should positively contribute to readiness in applying big data as proven by similar studies [29,31]. Data security issues are of particular concern to organizations when adopting big data [33]. Big data includes a lot of personal information and with rapidly increasing capacity, this is considered a valuable data source which may be exploited by unrelated third parties or cybercriminals [33]. Similarly, Motau and Kalema [17] highlighted that it is a precondition factor affecting the readiness to analyze big data. Adopting big data to keep up with partners and maintaining the firm’s internal balance with them is a key readiness marker[15,16]. Some empirical research studies have suggested that trading partner pressure is an important determinant for IT adoption and use [34]. Financial resources refer to the enterprise's budget to invest in information technology infrastructure systems, investment in training high-quality human resources capable of analyzing big data, and financial resources to maintain the operating system when the company implements big data deployment. The financial resource is an important factor that influences the application of big data [15]. 6
Predicting the factors that influence the readiness to adopt big data in SMEs is an important contribution to help managers better understand the factors affecting readiness to adopt big data in their enterprise. Hence, managers could build a big data adoption strategy that is right for their business. It will help SMEs increase revenue and consequently increase the contribution to the state budget. This research study is also a valuable reference for government agencies, for example, the Vietnam Association of Small and Medium Enterprises and the Ministry of Investment Planning to encourage the future implementation of big data by enterprises. In addition, service providers also could understand the factors that affect the readiness of enterprises to apply big data thereby also developing a strategy to provide suitable products for businesses. This study is useful for the implementation of BDA by SMEs in Vietnam as well as SMEs in some developing and underdeveloped countries. 10. 11. 12. 13. 14. 15. 16. 17. 18. 20. 21.