Predicting the Success Rate of Entrepreneurship in Biotechnological Companies Using Machine Learning (Case Study: Iranian Companies)

Abstract

As a newly emerged technology, biotechnology is developed by cellular and bio-molecular processes to provide products that help improve the lives of humans and other living. The bio-entrepreneurship concepts have targeted scientists with feasible business ideas in the biotechnology field, considering the competitive environment in the biotechnological industry. Therefore, they should be able to determine the success of their company by the operational and organizational factors. This study tried to identify the success rate of biotechnological companies by having the situation of their organizational and operational factors. These factors were (a) having a clear understanding of the market, (b) expert human resources, (c) the ability to take risks, (d) organizational structure, and (e) strategy and leadership. In this situation, machine learning, which is based on learning the observations by a mathematical algorithm to predict the outputs of a new unrecognized sample, can be helpful. In statistical regression models, the dependent variable is calculated through an equation using the given input features of samples. The number of input features does not typically exceed 2 or 3. In contrast, machine learning models can learn a database, including hundreds of input features and corresponding dependent variables or targets. 120 Iranian companies in the biotechnology industry were provided with a questionnaire, and the obtained data were trained to a support vector regression algorithm to provide a reliable dataset to train the learner algorithm. Various kernel types, i.e., linear, polynomial, Gaussian, and sigmoid, were considered during the design of the algorithm. The created machine included five input factors (having a clear understanding of the market, expert human resources, the ability to take risks, organizational structure, and strategy and leadership) and one output (a success rate between 0 and 10 defined by the CEO of the company based on net revenue and other economic criteria). Machine learning is required for tasks that are too complex for humans to implement directly. The model used in this study had five input variables, which made it impossible to provide a statistical model. Therefore, some tasks are so complex that it is impractical, if not impossible, for humans to explicitly work out all of the nuances and code for them. Instead, the machine learning algorithm was provided with a large amount of data to explore the data and search for a model. The results showed that a support vector machine with the Gaussian kernel resulted in a coefficient of determination (R2) and meant squared error (MSE) of 0.91 and 0.011, respectively. Linear and polynomial kernels resulted in slightly lower performance. Moreover, a sensitivity analysis was performed to determine the most influential factor in the success of biotechnological companies in Iran. The most significant impacts on bio-entrepreneurship's success were expert human resources, a clear understanding of the market, and the ability to take risks. The authorities should consider this since reports have shown that one of the remarkable limitations in developing bio-entrepreneurial firms in Iran is their managers' lack of business skills since they are biological scientists with poor management skills. The findings of this study show that by having the values of the five input above variables, the success of bio-entrepreneurial companies can be predicted with acceptable performance.