Data Analytics 2.docx PDF

How many times have you heard, \"Trust me, I have a lot of experience. I already faced this situation many times. I know how things are going\"? This is like a famous line, but it\'s crucial to remember: don\'t forget the human element. A good manager, a good leader, can rely on data without forgett...

How many times have you heard, \"Trust me, I have a lot of experience. I already faced this situation many times. I know how things are going\"? This is like a famous line, but it\'s crucial to remember: don\'t forget the human element. A good manager, a good leader, can rely on data without forgetting the human aspect. So, W. Edwards Deming is often credited with playing a key role in the revival of the Japanese economy and companies like Toyota, which adopted the principles of Deming. The philosophy of Deming, referred to as the Deming Management Method, emphasizes the importance of focusing on process improvement rather than blaming individuals for problems, promoting teamwork and collaboration, and seeking continuous improvement. These are the values of Deming. A quote from Deming is, \"In God we trust; all others must bring data.\" So today, we refer to the Deming method to improve processes instead of blaming individuals for problems. For that reason, don\'t forget the human element; we can work with data while respecting the human aspect. Here, I want to introduce the concept of the HIPPO---it\'s an acronym for the \"Highest Paid Person\'s Opinion.\" It refers to situations where decisions are made based solely on the opinion of the most senior or highest-paid person in a group, regardless of whether that opinion is the most informed or well-supported. This can be a problem because it can lead to lower-level employees or young managers feeling discouraged from sharing their ideas, feeling undervalued, and disengaged. To avoid the HIPPO effect, it\'s important to encourage a culture of data-driven decision-making based on evidence and analysis rather than solely on individuals\' opinions. HIPPO is not alone; it has some friends like the Rhino (Really Here in Name Only) and the Zebra (Zero Evidence But Really Arrogant). The concept is to be aware of relying too much on opinions without data. As a manager, it\'s important to be always present in the company, to be transparent, and as clear as possible. Just some tips and advice for your career as a manager. So, we start to talk about data-driven strategies and why it\'s important to define a data-driven strategy. First of all, how is it possible to foster a data culture? What does a data-driven culture look like? We have eight common traits of an organization powered by data: 1\. Everyone embraces data, starting from the executive suite. It\'s important because all changes must start from the top of the company. 2\. Data uncovers opportunities to drive improvements. 3\. New ideas are tested. 4\. Employees\' data skills are prioritized and developed. 5\. Data is accessible and consistent. 6\. Eliminating silos is important---to eliminate silos as much as possible from both technological and social points of view. 7\. Data is a business asset, delivering competitive advantages. 8\. Ethics and privacy are central tenets of data use. Don\'t forget the human element. Behind the data, there are always humans. We produce data, and if data is the new oil, we represent the oil for new companies. For that reason, don\'t forget humans, thinking about ethics, privacy, transparency, and so forth. Talking about data, now, how to create a data culture? In order to answer how to create a data culture, I used the definitions of four principal consulting firms worldwide, starting with McKinsey. Do you know McKinsey? So, what does McKinsey say regarding how to create a data culture? First of all, it\'s not possible to import culture, and you cannot impose culture. The goal is to make better decisions by building and supporting decision systems. Commitment from the CIO and board is crucial; changes start from the top of a company. Data democratization is important---data must be accessible with the right level of privacy, security, and protection across the entire company. Always think about an ecosystem and not a single platform. Eliminating silos and developing risk management are necessary. We need to understand the risks. Change agents are effective, often more experienced professionals with a deep understanding of the business, who can be more effective in driving change. The experience is always important, but today it\'s important to rely on data. This is the concept of change agents for McKinsey. Integrate the right talents---young people, for sure, and managers as well. Share as much as possible. Now, we move from the company to Harvard Business School, and you can find a lot of similarities. Data becomes the universal basis for decision-making. Since we started the lesson, we can define this phrase as commonplace: a data-driven culture starts at the top of the company---from the CIO, the president, the board, and so forth. We are talking about how we need to make decisions based on data. But how do you prevent someone in the company from manipulating the data to make their idea look better? At the end of the day, you are dealing with people presenting data to one another with their own biases and selfish goals. For that reason, it\'s important not to pigeonhole your data scientists. We must rely on data, collaborating with data scientists who support us, but without isolating them. As a manager with a clear vision about data culture and a data-driven company, it\'s important that if the data are not in line with our expected perspective and our ideas, it\'s not useful to manipulate data to align with our expectations. So, it\'s important to rely on data as it is, without manipulating it. I believe that data is read differently from individual to individual. So you need to be updated and know what results you see from the data. Remember that information is related to the context; your context could be different from mine. So we have two different levels of information starting from the same data. Going ahead, fix basic data access issues quickly. Quantify uncertainty if possible. Make proofs of concept simple and robust, not fancy and brittle. I prefer to start with simple proofs of concept and then extend step by step by project. In this case, it\'s possible to build an ecosystem block by block. Training is offered just in time. It\'s important to be aware. Use analytics to help employees, not just customers. For example, the churn rate is a metric used to describe the rate of customers who are leaving or at risk of leaving a company\'s service or goods. There is another concept similar to churn called attrition rate, which refers to employees. In this case, we are using analytics to help employees and not just customers. Another consulting firm that I used to explain how to create a data culture is MIT Sloan School. Connect data to organizational culture for both social and technological points of view. Data professionals should speak the language of business. Experts in the domains are important; we need the expertise. The relationship between data and business operations is crucial. Correct the perception that data is just for techies. Remember that data democratization is the first concern. Show how data solves problems at all levels of the organization. Today, from a system architecture point of view, flexibility can be challenging. But this is the world we are living in. For that reason, we need to be more flexible if possible, working step by step, and think about ecosystems and not single plans. If you start with a project from the technological point of view that involves your entire company, it\'s something impossible to achieve in the short term. So it\'s better to split the problem, working with small projects like proofs of concept, and then it\'s possible to achieve consistency through flexibility. The last one is from Forbes. How to create the culture? Data is an essential part of business processes. This is a data culture test. If your company leans on the left side of the scale, you are more data-detached than data-driven. It doesn\'t mean you don\'t have data or don\'t use data regularly, but it means that data is not essential to how you operate today. Most organizations, small or large, have huge amounts of data. But its presence doesn\'t mean it\'s actively relied upon for decision-making. If you don\'t have a data-driven culture, you need to foster it. Unfortunately, data-driven culture cannot be purchased or manufactured; it must be developed and reinforced. If you want to foster a data-centric environment, you want to focus on these four pillars: 1\. People: Shifting the collective mindset of your people to embrace data is one of the hardest challenges. You need to be diligent and patient as you attempt to steer your team in a new direction. 2\. Skills: To succeed with data, your employees will need specific data-related knowledge and skills. 3\. Tools: Over time, organizations can accumulate a variety of data systems and tools. This can be a nightmare. For that reason, it\'s important to sharpen the toolset. 4\. Data Set: Solidify the data set. The relevance and quality of data will determine whether it is embraced by people within your organization. This is how Forbes defines the strategy to create a data culture. As you can see, there are a lot of similarities with the other consulting firms. The main concept behind a data-driven strategy is the usage or sharing of data in order to achieve goals consistent with the business strategy. Always consistent with the business strategy, develop data governance---rules on data management. Implement the right level of data management, including lifecycle management and data enhancement. I think that to adopt a data architecture, there should also be a change in the organization. This change must start from the top of the company. For example, giving the right importance to data architects, data science, and data security. You are talking about a strategy managed by top management, built by architects and security, and managed by data scientists and data engineers. Your sentence is perfectly consistent with this. Going ahead, starting with data, there is a concept that we must not forget---it is the FAIR principles. FAIR is an acronym that stands for Findable, Accessible, Interoperable, and Reusable. The FAIR principles, from a technical standpoint, are related to the data-driven strategy. Always use FAIR principles when we work on data. Going ahead with the data strategy, instead of giving you the definition directly, we start from some questions. The main questions behind the strategy are: How can we use data for generating value for the business? What existing data do we have? Do we govern our data? To build a robust data strategy: 1\. Identify business objectives and define related data strategy objectives. It\'s important to start a project from the business objectives and not from technological choices. If you want to build data-driven decision-making processes, you need to start by identifying business objectives. 2\. Conduct an assessment of the state of policies, procedures, processes, and so forth. 3\. Define a roadmap to target the set objectives, including incrementally and iteratively. 4\. Identify and implement necessary change management actions to support changes in organizational culture, technological tools, and business processes. For that reason, it\'s important to identify the business objectives. As Peter Drucker said, \"There is nothing quite so useless as doing with great efficiency something that should not be done at all.\" So, first, define business objectives, and then we can start to talk. Moving to data governance, we have similar questions: Who is the data owner, and who has the right to access the data? What security measures are protecting the data? Which reliable data sources are used? Data policies, quality, risk management, compliance, business policies---all are part of data governance. Remember that when you start a new process or project, you must integrate it with all business processes. You cannot consider it as another platform. Always think about an ecosystem and not silos. Getting the balance right in terms of doing things right and doing the right thing is crucial. The last one is related to data management and data enhancement. Here we are moving into the technological standpoint: Which methods and technologies are used to acquire or store data? Which tools are used to ensure data quality? How is the integration of different systems and data sources ensured? What measures are taken to ensure data accessibility and reuse? Planning, documentation, and monitoring of data management define rules and responsibilities, contact information, how data is acquired, and processes. We need to have the full lineage of the data from gathering to the result. Without having the lineage of the data, it\'s not possible to build the lifecycle, and we cannot be sure that our data is reliable. From the conjunction of strategy, governance, and management---the company vision about data and culture---we can map the data maturity. Here we have three levels of data maturity: 1\. Data-Aware: Organizations recognize the importance of data but aren\'t yet leveraging it effectively. 2\. Data-Proficient: The organization becomes more competent in its data handling and starts to integrate data analytics into its operations. 3\. Data-Driven: Data is deeply embedded in the company\'s culture, with strategic and operational decisions made based on data insights rather than intuition. This data maturity framework illustrates the stages an organization goes through as it establishes its use of data. The maturity journey typically spans from simply recognizing data\'s presence within the organization to fully leveraging it as a strategic asset that guides every level of the company. And here we have a list of the main reasons for failure. Consider the seven reasons why companies fail to become data-driven. The first one is viewing data-driven transformation as something that impacts only the departments. That\'s right. So, digital transformation starts from the top of the company and is not just any department\'s responsibility; it is the responsibility of the entire company, not only for data management and from the technical standpoint, okay? Focusing on the data. Focusing on the data rather than the strategic and business objectives. So, as I said before, it\'s important to define the business objectives first, and then we can focus on data, okay? It\'s not starting from the data that we can become a data-driven company. Another important problem is related to the cultural shift. So, digital transformation is not just a technological change; it\'s also a cultural shift. Businesses need to foster cultural innovation and experimentation to embrace new technologies and adapt to new ways of working. And it\'s important to spread the concept that all the new technologies\... So, how to say, in a polite way, it\'s not just about replacing manual tasks with computers, but really about using technology to create new opportunities and improve efficiency across all aspects of the business. So, in a nutshell, buying a new computer, installing new software, and, I don\'t know, a new agreement with Google Analytics, is not the solution and is not the way to become a data-driven company. We need a cultural shift. This is why it\'s so important for the supreme results that I mentioned. Organizations need to adapt to change differently. We need a new mindset: fail fast, learn faster. So failure is the foundation of innovation for sure, and focus on the long term. So becoming data-driven is a process. Success is achieved iteratively. This is the problem. So this is the main point: step by step, block by block. Okay, questions? Is it okay? Can I go ahead? Great. Okay. Great. Okay. Now let me go a bit faster. So, always in terms of data-driven organizations, just to have other elements: IT or not IT, that is the question for sure. And as we said before, it\'s not just an IT question, but we need the organization. So we need a data-driven organization, starting from the CIO and then all the board, all the C-levels of the company. In a data-driven organization, we have a new set of roles. We have, for example, a CDO---Chief Data Officer---in order to manage and develop data as a strategic business asset, and many other responsibilities in your organization. If we need to eliminate silos from the technological standpoint, we need to do the same for the organization. So build a platform, eliminating silos from the technological standpoint, and at the same time eliminate organizational silos. So, work together. So this is the concept. Okay, we need to work together and break silos in groups, okay? Something like the agile methodology. Okay, so be transparent, share, work together, and so forth. You need to build a data scientist team. It\'s important. Here, we have a suggestion on how to build a data scientist team. So in order to have a data-driven organization, you need a data scientist team. Here we have it. After that, we need to implement some framework in order to have data science processes in your company. So it\'s important to define a framework. This is one of the many frameworks that we have in order to apply and improve the data science process. In this case, the OSEMN framework: Obtain, Scrub, Explore, Model, and iNterpret. So it\'s just related to the project lifecycle and not the business objectives, okay? So here we mean business objectives, and then remember that you can\'t improve what you don\'t measure, and then manage. Another concept in the data-driven strategy and data-driven organization is related to process mining. So, what is process mining? Process mining is a new science, a new field of science where we have different data sciences, everything related to data mining and BPM---Business Process Management. Oops, sorry. Process mining is a science that helps organizations to optimize all the organizational processes, the processes, and all the processes involved in the business line. For example, I don\'t know, if your company is a pizzeria, so ordering a pizza at the counter shop is a process; ordering a pizza via the website is a process. If your company sells, I don\'t know\... \"Is the lack of culture due to awareness?\" \"Yes, for sure. If you don\'t have enough culture, it\'s impossible to be aware of the data-driven organization, for sure.\" And so, process mining, yes. So, yeah, where were we? I lost track. Okay, so the goal is to improve the processes---internal processes---in order to identify, for example, bottlenecks, loops, inefficiencies, in order to understand if it\'s possible to implement, for example, robotic process automation. So, and totally based on data, process mining uses just one kind of data in order to produce the first level of results, which is represented by the logs of the software, so all the applications. So going to the example of the pizzeria, in order to apply process mining in a pizzeria, we need all the logs that come from the software for ordering a pizza, where you have all the steps for ordering a pizza. So starting from the order, I don\'t know, preparing the pizza, baking the pizza, delivering the pizza, and payment, okay? So all the processes end to end, and all the logs that the software produces must be used by the process mining software in order to improve your processes. Going ahead, we have the concept related to security, protection, and ethics. But about ethics, protection, and security, we will talk in the next lecture in depth. So, let me jump to the concept that is important and useful for your exercise. So here we have the correct\... Oops, there is a problem. Okay. Here we have a process. What do we have? Analyze data, extract valuable information, identify business needs, and make decisions. That could be a good approach, but it is not a data-driven decision-making approach. In order to have the DDDM approach, first of all, we need to identify business needs, then acquire data, after that analyze data, and at last make decisions, okay? So this is the right process in order to apply DDDM. These are the correct steps. Jumping to the end of the presentation---the other topics we will see in the next lecture---and I go here to the business case that is your homework. So let me be honest, it\'s not an assignment, and there is no evaluation, no score. It\'s just an exercise. It\'s almost for you in order to understand if the data-driven strategy that today is clear for you, okay? So you can work alone or in groups, whatever you want. You can organize by yourselves as you prefer. And I will ask you during the first hour of the next lecture to show the results, okay? So you don\'t need to prepare a PowerPoint document, a written report---simple, just bullet points with the answers. Okay, so really, really simple. And let me repeat, it\'s not an assignment, and we will not give you a score, okay? It\'s just to start the discussion in the next lecture and to share the approach. Okay, guys, is it clear? Great. So Jane says, during the instruction process, if we realize that we don\'t have sufficient data or whatever, this is a good question, Jane. This is a good question. So if your company doesn\'t have enough data, or the quality is not at the right level, you need to improve your data capability, you need to improve your data. You need to add or enrich your data if it\'s possible. And in the homework, we will talk about it. For example, if your company can implement a loyalty program in order to retrieve information from their customers, for example. Or there are a lot of companies that sell data. So it\'s a new business model that\'s called DaaP---Data as a Product. You can purchase data from other companies, okay? For example, one of the biggest companies is called Talkwalker. Talkwalker sells data that comes from social media, okay? You can buy---you can purchase data from them as well, or there are many other companies. Okay, Jane. Abdul Salam. Yeah, I am about to explain the homework. That is really, really simple. So, let me go back here. So imagine being a retailer with 30 stores all over Europe, with an annual revenue of 100 million euros, and you have data on your customers, products, inventories, sales are very short, and so forth. You can add---you can have all the data you need; it\'s not a problem. So the challenge is to implement an effective decision-making process using customer behaviors and product performance data to develop a data-driven strategy. So here I typed, working in teams, list, define. So I understand that you come from different master\'s programs, and you are all online, so you are not in the same place. So if you want to work in teams, fine. If you want to work alone, for me, it\'s the same, okay? So no problem. So please define at least five business needs that the DDDM process is expected to answer. Okay, five questions. Let me go back to the presentation. Where is it? Okay. So this is the DDDM process: identify business needs, acquire data, extract valuable information, make decisions. Okay? So you have the presentation, so you can use it. In fact, define the main steps to set up and implement an effective DDDM process, and at least three outputs---KPIs, reports, graphs---it\'s up to you, it\'s your choice, related to the business needs that need to be evaluated for making effective decisions. Okay? Is it clear? Yes? Okay. Okay, let\'s go back here. Okay, so where can I find the data? No, you don\'t need the data, Simon. It\'s just, how to say, it\'s just an exercise. So in order to try to describe the steps, okay? It\'s not\... okay. So it\'s really, really\... yeah. It\'s a theoretical exercise? Yes, for sure. Yeah. And now I\'m going to share with you again the presentation. Let me use WeTransfer. Okay, are you ready? I shared it at the beginning of the lecture. Okay, here you have the WeTransfer link. Okay, no, no, Simon, you don\'t need to send anything. So, of this lecture\... okay, during the part of this lecture\... the date is next Monday, if I\'m not wrong. Let me check the calendar. Where is it? Yes, next Monday. Yes, so 4th of November. Okay, yes. We will discuss together about your results, okay? So it\'s not just to say, without scoring---just a way to discuss together about how to approach a DDDM process. Okay. Is it clear? So did I already share the material? Mariano, can you show the previous slide, please? The DDDM, Mariano. Simon, and we need\... Okay, thank you so much, guys. So Monday, could you provide us some references to expand our information about the subject? So, there are a lot of\... so during the\... there are a lot of URL links. Yes, I saw the email. Yes, yes, you can. Yes. Yes. If you want some references, so in terms of books, or you want to study on\... I will give you in the next lecture a list of, for sure, yes. I will note it down in order not to forget. Okay. Okay, guys. Let me go back to the previous slide where we have\... okay? So this is the challenge. Okay, that\'s all for today. Many, many thanks. I hope that you enjoyed the lesson, and it\'s the\... the idea is to provide data to make a decision. We should consider that the business needs come---so the business needs come, yes, for sure. Yes. Yes. From C-level. Your attention? Yes, yeah, you\'re right. So this means by C-level\... C-level is the senior level, is the top-level staff, managerial companies, okay? So top manager---it\'s called the C-level. C stands for Chief. Okay. Okay, guys. Many, many thanks. Thank you for your time. Thank you for your patience. And I wish you a good evening, and see you. Thank you, Richard. Thank you. See you next Monday. Thank you. Bye. All right. Thank you.