AHBETE

[miniMBA_05] Leveraging Disruptive Technology in the Digital World

The Internet of Things

• Represented by a shift from humans to computing devices as the focus of connection and communication of data to enhance personal and business processes
• Potentially billions of devices that could connect – multiple devices for some people.
• Interconnected to machine learning and artificial intelligence as much of the data that drives those fields is going to come from the data being collected on the Internet of Things.
• Needs to balance the goal of creating technologies that will enable more sustainability because of reduction in waste and reduction in energy consumption with the trade-offs of higher energy consumption, contribution to the carbon gas problem, and e-waste from all the devices.

Benefits

• Communication – e.g. smart digital meters that can capture electricity consumption on a moment-to-moment basis rather than a monthly basis.
• Habit Changes – e.g. control devices on air conditioning systems that can smooth out the consumption process and production processes while increasing efficiency and providing savings to customers.

Levels at which Internet of Things (IoT) devices typically operate

• Individual - smartwatches, fitness trackers, smartphones that are able to monitor our movements and activity, certain health characteristics, and our immediate environment
• Household - smart appliances, smart thermostats, lighting systems, security systems, etc., that allow the household to be able to monitor and manage the environment more effectively
• Organization - industrial sensor networks, inventory, tracking and sensing technologies that enable organizations to monitor and manage their business environments
• Community - smart street light systems, traffic cameras, waste management sensors that can be used to make communities more efficient

Potential Challenges

• Security - more connection to the internet means more avenues for cyber criminals to leverage, especially if security is not a core functionality of the technology.
• Privacy – benefits of personalizing our connected devices (e.g. fitness tracker) and the resulting data that comes with potential risk of that date being shared (e.g. health data to insurance company) which could result in higher costs

Artificial Intelligence and Machine Learning

What are machine learning and artificial intelligence and what’s the difference?

• Intelligence - acquiring information and knowledge and being able to transmit that to others.
• Artificial intelligence (AI) - the ability to incorporate human-like intelligence into machines and computers.
• Machine learning - a branch of AI; providing data to machines, giving them the ability to learn by themselves from experience

In classic AI, also called symbolic AI, computer scientists try to use logic or other forms of math to let machines do the reasoning and solve logical problems.

In modern AI, very large amounts of data can drive all the reasoning and solve problems. Machine learning has become a very important branch of modern AI.

Artificial intelligence technology, from a business perspective, can be divided into three groups:

1. Assisted intelligence - widespread machine learning data-mining apps that business professionals employ to collect and analyze data, such as stock market predictions.
2. Augmented intelligence - current stage of our AI development where both people and machines, or the algorithms, make collaborative decisions
3. Autonomous intelligence - potential in the future for AI robots or agents to make decisions for humans

Why do machine learning and artificial intelligence matter to business?

• They can accelerate the facets of a business by going beyond typical automation that manually replaces repetitive tasks with intelligent automation that’s learning how to gain efficiency on its own.
• They introduce new products and new ways of being able to apply artificial intelligence such as self-driving vehicles or tools that work alongside humans such as robotics.
• They allow businesses the ability to customize more to their customers such as the ability to target specific customers with specific ads in various platforms or social media.

What are the professional roles in an analytics ecosystem?

• Technology provider - computer scientists, statisticians, data scientists who have very deep understanding of the algorithm, the math, the statistical models. They write apps or develop new tools for the different business units. Not every business needs this level of professional.
• Analytic accelerator – trained professionals who know the coding but are not doing the coding. They have a deep understanding in both the business side and also technology side. They are more commonly needed and channel the business question into the technology side and translate, or tell the story, from the technology side to the business side.
• General users or user organization - Business professional who need to be exposed to new technologies through tools which can be used without any coding background

Functional knowledge - Business professionals don’t need to develop the algorithm in order to understand how it works. Leaders need to understand the capabilities and understand their business well enough to know where to integrate those latest technologies into their business process and operations.

Risks

• Privacy - need to safely harness all that data to ensure customers, employees, and suppliers are protected, such as the protection provided by HIPAA and FERPA.
• Bias - need to ensure that bias is monitored and course-corrected as needed since artificial intelligence is created by humans and we inherently have bias. If there is bias in the data, it will translate into bias in the machine learning and artificial intelligence models
• Ethical - ensuring that AI is being used as a tool for important and ethical purposes

Challenges

• How to distribute the wealth AI creates? As a business leader, we should offer answers to our government, to our community, and to our people.
• How to retrain a workforce? The goal is a very close integration between the technology and the existing business, so the current workforce with deep understanding of operations and customers will be a key asset. One solution is better collaboration between higher education and companies to create new ways to retrain our current workforce with the latest technology.

Cyber Security

Technology is just one component of cyber security. The entire cyber security domain includes people and processes in addition to technology.

People

• Security breaches tend to be more of a people-related problem as opposed to technology. Employees need to understand the security threats that they are going to face.
• Organizations need to have a culture of security where employees have the necessary training and skill set.

Processes

• Organizations need to have a process framework that is designed to align the security challenges of the company to the technology.
• An inefficient process, such as a poor employee termination process, could contribute to security problems.

Managers should institute policies and in-house programs that can potentially mitigate cybersecurity threats:

• Conduct a very thorough risk assessment to be able to understand the inherent risk associated with the type of product and with the type of process, such as multiple partners in the supply chain.
• Anchor employee training around the risks identified.
• Create a strong security culture with policies, end-user agreement, and user policies that capture the emerging security challenges.

Tips for Creating a More Secure Organization

• Segment your network - a potential breach will have less exposure to the network.
• Implement a layered approach to security - it is important to go beyond perimeter security so that your core servers are in more secure places that have more countermeasures such as intrusion detection systems.
• Leverage your access control - give people access only to data that is needed for daily job responsibilities.

No industry or sector is immune although some, such as healthcare and tech sectors, have recently been more vulnerable to cyber security threats.

Individuals can balance the convenience of using technology with privacy concerns and cyber threats by being aware and trained to implement some of the control mechanisms that can diminish risk.

Organizations need to look at the cost-benefit. New technologies can help change the market dynamics and make processes more efficient but the appropriate control mechanisms must not be neglected. Cyber threats are real and they are continuously evolving. Organizations have to be dynamic and agile in their ability to have adequate security mechanisms in place.

Analytics and Data-Driven Decision-Making

• Using data models and algorithms to improve strategic decisions that underlie core business operations.
• Plays a central role in today’s business environment, especially as a means for firms to gain competitive advantage by using insights from data that they collect
• Three categories of analytics are descriptive, predictive and prescriptive

Descriptive Analytics

• Answers the question “What is happening right now?” or “What has happened in the past?”
• Used as a base for the other two types of analytics
• Applications: dashboards showing current data and trends

Predictive Analytics

• The process of using historical data to make predictions about future events
• Applications: credit scoring, direct marketing, churn models, predicting healthcare outcomes
• Predictive models can be constructed using structured or unstructured data, such as text data generated from social media postings or product reviews.
• Forecasting, or time series, models are used in almost every industry are based solely on the temporal behavior of a variable. An example includes forecasting demand for products at certain times of the year.
• Data mining can be thought of as an umbrella term that encompasses predictive analytics. It is unsupervised learning and works by finding similar observations and placing them into groups to create things in marketing like a customer profile, which then can be used to better target certain groups of people.

Prescriptive Analytics

• Finding the best, or most optimal, course of action for a particular course of action based on a number of complex factors
• Factors could include finite resources, available labor hours, investment capital, supply chain restrictions, etc.
• Applications: logistics, supply chain, healthcare, petrol chemical industry, finance, military
• Other techniques that fall into the category of prescriptive analytics include design of experiments, simulations, and other decision analysis methods such as game theory.

Do descriptive, predictive and prescriptive analytics operate independently?

They are intricately linked and regularly complement one another in analysis.

• A predictive model often represents an objective function in a prescriptive model since it can predict the outcome of different decisions.
• Descriptive methods enable us to make more realistic models because they help to properly estimate the parameters that go into the models and ultimately dictate their accuracy and effectiveness.

What are the skill sets necessary to be successful in analytics?

Necessary technical skills include:

• Coding
• Database management
• Basic statistics in math
• Modeling and simulation

Possess a working knowledge of:

• Statistics
• Computer science
• Mathematics

From a business perspective, it is also critical to understand, interpret, and communicate results with stakeholders that may or may not have the same skill set that an analyst does.

Communicating with Data

• Start with the right question - a question that solves an important problem and that is going to bring value to your business
• Make sure you have the right data to answer the question
• Develop the right solution – a clear and transparent analytic solution

Reproducibility in data analysis is a quality standard that describes an analysis that is clear, well-documented, and open. An analysis is reproducible if others can reproduce the analysis and obtain the same results.

Solving Big Complex Problems

• Solve a simpler version of the complex problem then continue to iterate, making it more and more complicated as you go, hopefully, getting to the problem that you want to solve.
• Take the problem and break it down into smaller pieces and to solve the individual chunks and then to build up to a bigger solution.

The Extraction, transformation and load data (ETL) process is used for creating a Data Warehouse and Data Mart to solve big complex problems.

1. Companies collect a huge amount of business data from different sources on a daily basis through internal sources such as ERP systems, corporate websites, mobile applications as well as external sources such as purchased data from credit card companies.
2. They extract and clean up each source of data, transform and load them to a gigantic central repository of integrated data called a data warehouse.
3. Next, they create subsets of the data warehouse, called data marts, for supporting specific business lines or teams such as sales, marketing, manufacturing, or financing.
4. All these daily efforts are directly related to iterative and reproducible processes of breaking down big and complex data into smaller and manageable ones.

How to Effectively Communicate Business Solutions

Focus on the audience to make sure that we transfer the knowledge effectively by tailoring the communication based on the following questions:

• What does our audience know in terms of the context of our data?
• How much technical knowledge does our audience have?

One method is to use the value proposition canvas from Strategyzer as a simple way to understand the audience for your solution. Although designed for entrepreneurs, this tool is useful for analytic solutions because it forces you to think about your solution from a value proposition standpoint. It can transform the way that you communicate with data. This value-driven approach ensures that you are communicating results that are important to your audience, and not just results that are important to you as the analyst.

Ways to Communicate Ethically in an Unbiased Way

• Carefully implement your visualization techniques. A change in how a visualization is structured can manipulate the data to be misleading.
• Use infographics appropriately:
• Include the baseline
• Be mindful of the unit of measure for y axis and x axis.
• Include all of the data, not a “cherry-picked” portion
• Keep with common conventions, such as a typical use of color.