**9.23 Social media**

Social networking companies such as Facebook have been using predictive analytics algorithms that enable updates regarding a brand to be available to the user after

a product has been "liked". Therefore, users are able to see posts, which improve their engagement rates with their individual networks such as posts that their friends have engaged with [16]. The programming language used is JavaScript due to its cloud computing feature and the ability to make changes to an already existing algorithm. The methodology used is the creation of an algorithm that enables the site to link a liked product to a user's page. The process includes the statistical analysis of a decision tree in which the website is automatically coded to link a "liked" product to the user's page. The final outcome is a user experience in which when a person likes a product, the updates regarding the product appear on their page in the future. This implies that Facebook will promote the ability of marketers to promote social engagement with customers.

### **9.24 Manufacturing**

In manufacturing companies, machine-learning algorithms have been used to understand the machine problems that are likely to be encountered in order to apply preventive practices to keep the supply chain operational. At Georgia Institute of Technology, machine-learning algorithms provide the opportunity to promote forecasting the likelihood of machine failures, thus, enabling the technicians to perform maintenance practices [3]. The machine learning language used is a C program with capabilities for creating codes that enable calculation of statistical tests such as regression analyses, linear regression, and multiple regressions. The methodology used is the creation of a computer algorithm in which past intervals of failures is added. The data are the failure times (the dependent variable) and the time interval (independent variable). A sub-program is created that enables the calculation of simple regression analysis, which establishes the relationship between machine failure times and the time interval. The preliminary results are the input values of failures of the machines against time interval. The outcome of the analysis is a regression coefficient, which can be multiplied by the current failure frequency to determine the next likelihood of the machine's failure. This ML algorithm has been applied in the performance of regular maintenance tasks on lathes, grinders, saws, and gears (**Figure 6**).

### **9.25 Government institutions**

In the United Kingdom (UK), the Ministry of Defense uses machine learning algorithms to explore and organize public documents. This is achieved by creating algorithms that enable the identification of documents depending on their subjects and conducts the analysis of information for the purpose of finding patterns and anomalies in data systems [25]. The algorithms are also implemented in the detection of fraudulent activities, transactions, or activities of any public official for personal gain. The algorithms have been effective in the detection of activities such as money laundering, the creation of counterfeit trade items or the duplication of business documents. The processes include the installation of machine learning languages into the systems of the organizations, the creation of computer programs, testing, and implementation [35]. The inputs are information regarding future activities such as the attempt to change the content of documents in order to achieve personal objectives or defraud the government. The program is capable of providing information about the perpetrators of the acts and insights on characteristics that can be used to trace them.

**Figure 6.**

*Summary of the improvements made at General Electric [3].*
