FBA Lecture 01 - Overview & Intro to Analysis PDF

BUSI4371: Foundational Business Analytics Instructor: Dr Evgeniya Lukinova Week 1: Overview & Intro to Analysis I. Overview Welcome to Foundational Business Analytics. Today, we will be having a gentle, rolling start into the course, and you will learn: 1. Housekee...

BUSI4371: Foundational Business Analytics Instructor: Dr Evgeniya Lukinova Week 1: Overview & Intro to Analysis I. Overview Welcome to Foundational Business Analytics. Today, we will be having a gentle, rolling start into the course, and you will learn: 1. Housekeeping 2. What is Modern Business Analytics? 3. Why is it important & where is it applied? 4. What tools/skills will you need to master? 5. What will be required of you in this course? 6. What is the road ahead. 1. A beginning: logistics Instructor: Dr Evgeniya Lukinova Course Hours: 11am - 1pm Tuesdays (Lecture) 11am – 1pm Wednesdays (Computing) Location: BSN A34 Email: [email protected] Teaching Support: TBA 1. A beginning: a bit more about me Dr Evgeniya Lukinova Room B78c, Business School (North) Assistant Professor in Behavioural Analytics N/LAB Interdisciplinary scholar with research focus on aggregated and individual human behaviour 1. A beginning: course structure Structure: The course has 11 weekly lecture sessions, accompanied by a weekly computing session. Most topics covered are set against real-world case study examples. 1. A beginning: your commitment Total Learning Time: 200 hours Contact Time: 44 hours (4 one-hour sessions per week) Background Study: 56 hours (at least 5 hours per week) Coursework Hours: 70 hours Revision Time: 30 hours (20 test / 10 quizzes) 1. A beginning: Your Assessment Assessment 1: In-class quizzes (best four out of five) - 20% Assessment 2: 1 hour in-class technical computer-test - 30% Assessment 3: Individual Coursework: 3000 words report on a Business Analytics Project - 50% Please watch the video on assessment structure on Moodle and address all remaining questions about assessment to the chatbot or discussion forum on Moodle! 1. A beginning: Your Assessment 50% is a pass (this is ok) C 60% is a merit (this is good) B 70% is a distinction (excellent) A 80% is a top distinction A+ 90% is publication quality 1. How can you Get Feedback for this Module? From Lecture Sessions: Actively participate, contribute and engage. Get feedback as your ideas develop with in-class tasks and feel free to ask questions! From Computing: Please make sure you ask questions early on and come prepared by completing previous practicals. From the Moodle discussion forum. From each other! Out-of-class Availability: online by appointment or face to face by appointment Moodle Moodle will be used for: Access to the lecture slides and lab activities (available at least 2 days before the lecture). Quiz preparation and submission. Submission of Python test and coursework. Providing Information and Announcements. Supporting your inquiries through a chatbot and discussion forum. Make sure you are enrolled now. The forum is the only way to ask non-personal questions. 1. How can you Give Feedback for this Module or MSc programme? Polls in- and out-class. End of module evaluation. Feel free to email me with suggestions at any time if something isn’t quite working. Use the MSc programmes Teams Page: ○ Post your feedback and queries ○ Monitor the posts, files and tabs associated with the page 1. What about GenAI? “As AI becomes more ubiquitous, education should prepare students to a) learn how to use these tools effectively; and b) how to exercise judgment about why, when, where, and under what circumstances they should be deployed.” Frederick M. Hess, Director of Education Policy Studies at the American Enterprise Institute 2. What is Modern Business Analytics? 2. What is Modern Business Analytics? “Data analytics continues to be in high demand. Overall, 71 percent of employers plan to place recent business school graduates into data analytics roles in 2018. Thirty-five percent of companies hired Master of Data Analytics graduates in 2017 and 52 percent of companies plan to hire them in 2018” 2018 Corporate Recruiters Survey Report (Graduate Management Admissions Council) 2. What is Modern Business Analytics? “Overall, nearly 9 in 10 corporate recruiters plan to hire Master of Data Analytics graduates in 2022 (86%), up from 73 percent of the same recruiters who actually hired them in 2021” 2022 Corporate Recruiters Survey Report (Graduate Management Admissions Council) 2. What is Modern Business Analytics? TASK I: Well let’s start with asking you what you think Business Analytics is today… Split into groups of 4 (ideally) Introduce each other, and elect a “speaker” Write down your background disciplines Together isolate 2 key elements/parts of “Modern Business Analytics”. Write them down, and when asked the speaker will announce your table’s members, backgrounds and suggestions. You have 5 minutes… 5 minutes 2. What is Modern Business Analytics? There is no single definition of “business analytics”. In fact, it is often easier to define what business analytics is NOT: It is not “business analysis” (at all anymore in fact) It is not just simply calculating “statistics” in excel/SPSS. It is not about just blindly applying fancy “machine learning” But there is some consensus on what Business Analytics IS…. 2. What is Modern Business Analytics? Business Intelligence Predictive Modelling Data Mining Big Data Management Decision Analysis 3. Where is it being applied? TASK 2: Think of some examples of where business analytics is currently being used to improve performance… Split into your groups of 4 again Elect a DIFFERENT “speaker” Together come up with a list of at least 2 examples of where “Modern Business Analytics” is being used to improve business performance. Be as specific as you possibly can. Pick companies if you want. You have 5 minutes… 5 minutes  3. Where is it being applied? Targeted Marketing New Product Introduction Predictive Services Segmentation Customer Profiling Uptake Prediction Store Location Customer Lifetime Value Collaborative Filtering Customer Classification Stock Prediction Churn Prediction Similarity Matching Product Lifecycles Identifying Demographic Change Golden Customers Basket Analysis New Product Development Recognizing Market Trends Life Trajectories 3. Where is it being applied? 4. What tools do you have already? I’m going to ask what tools you have already and what tools you think you will need, and let’s see what we come up with… 4. What tools will we need? Traditional Business Analytics Software …are no longer good enough to meet our needs. -> WHY? Because Modern Business Analytics Requires use of: Scientific Methods  How good is the About finding out whether the About finding patterns that pattern at data fits a pattern. best fit the data... predicting new data. Hypothesis TRADITIONAL SCIENCE Generating Science DATA SCIENCE 4. Why do you need to think like a Scientist? Well, science is no longer about “description” It is just about finding the models which best predict data. That model might not be “true”… … but it is your best guess at the way the world works. … and our best option to therefore use when making real-world decisions. Thus, making it key for business performance. Prediction is core Models are the key 5. Models in Business Analytics Model - an abstraction or representation of a real system, idea, or object; Often a simplification of the real thing; Captures the most important features; Can be a written or verbal description, a visual representation, a mathematical formula, or a spreadsheet. 6. What tools will we need? So to do business analytics nowadays, you must have skills in the tools that can make predictive models easier to put together. This is the core of business analytics. 6. What skills will you need? Technical Knowledge Understanding Business Decisions Visualization/Communication Skills Statistical Modelling 6a. Intellectual Skills You must learn to be an inquisitive person. Intellectual curiosity is the key to success in analytics. Knowing the WHYs and HOWs of any business situation. You need to demonstrate a keen interest in understanding the core issues of a case study problem and working towards the specific technical solutions. These skills will be learnt – they may not be immediate. 6b. Quantitative & Technical Skills You MUST become technically oriented to do well. This is not the course for you if you’re scared of the technical side of things. Besides quantitative skills, you must become a logical and analytical person, with strong interpretation skills. You need to assess the utility of the data and make business strategies on the basis of your data. 6c. Communication Skills Once you gather, analyze and interpret data, the next critical step is communicating your insights to decision makers. You need strong verbal and written communications skills in order to convince some higher management or board of directors about the positive impact of your findings on the profitability of a business. We will assess your ability to translate technical analysis into business recommendations. 6d. “Self-Motivation” Skills You need to be a proactive person to do business analytics. You must take the initiative with this module to succeed. Besides the above four key skills, you also need to possess attention-to-detail, ability to see the big picture, team-working skills, and willingness to learn. You need continuous professional development and must learn advanced methods. 7. What will be required of you in this course? You will get much support but this course will demand a lot of you. Remember you are expected to do at least 5 hours work outside of contact time. If you do not do this, you may fall behind. We will direct you to learn the exact skills you need in the wild (the ones our partner companies have told us they are looking for)… Quick note about plagiarism All submitted work must be 100% your own. You may chat, collaborate and share ideas with each other… …but if any of your final submission looks like someone else’s, or if there is any suspicion your work is not your own you will be interviewed. If plagiarism is still suspected after interview you will immediately fail the module. Your case will then be forwarded on to academic complaints for further action. PLEASE DO NOT DO THIS! Quick note about AI (e.g., ChatGPT) The university considers the unauthorised use of AI tools false authorship: https://www.nottingham.ac.uk/currentstudent s/news/using-ai-tools-in-your-studies?dm_i=5 IL5,VE8H,4RN28D,3UTTW,1 It will be clearly communicated to you whether you may use AI tools in assessment and how you are permitted to do so. PLEASE ADHERE! 8. The road ahead Schedule: Week 1. Overview / Intro to Analysis Intro to Python Week 2. Fundamental Business Stats Flow & Data Structures Week 3. Making Linear Predictions (Quiz 1) Data Structures II Week 4. Classification Models Functions Week 5. Decision Trees (Quiz 2) Pandas Week 6. Node Impurity & Entropy More Pandas Week 7. Success & Ensemble Methods (Quiz 3) Revision & Sklearn I Week 8. Naïve Bayes & kNN Python Test (20th Nov) Week 9. Clustering I (Quiz 4) Plotting & Sklearn II Week 10. Clustering II Bringing it all Together! Week 11. Dimensionality reduction (Quiz 5) Revision II. An introduction to analysis We will start to be less gentle, and consider an example business analytics problem used in interviews. You will learn: 1. How to consider a ‘simple’ business problem analytically. 2. To think about the assumptions we must make in attacking it. 3. To consider concepts like “customer lifetime value” in analysis. 4. To see how valuable automating analysis can be. 5. To consider the jump that “modern business analytics” makes. task 1. Exploratory case study In this task we are going to do some business analysis and try and apply some of your existing knowledge to a case study scenario. From this we aim to explore what lines of attack are specific to “traditional analysis”… …and what items we need to learn to be able to achieve “modern business analytics”. Along the way we’ll also emphasize the need to think both precisely and systematically about the business task. task 1. Financial aggregation website Capital One is a credit/market intelligence company - but it is always looking for new sources of revenue. To this end, imagine we’ve proposed to them a subscription-based financial aggregation website. The site will bring together streams of data from within Capital One, and external partners. The aim is to use the site as a new revenue stream for current customers, and to attract new ones. task 1a. Financial aggregation website Research shows the company can charge £24/year in subscription for each customer using the service. Additionally the company can obtain £1/month from advertising revenue per customer. It will cost £600k year to obtain data for the site. - What is expected revenue per customer in a year? - What is expected profit per customer in a year? - What further knowledge might we require about this venture to answer that question in more detail? task 1a. Financial aggregation website - What is expected revenue and profit per customer in a year? Revenue per customer/year is: £24 + (12 ✕ £1) = £36 So profit per customer/year is: £36 – (£600,000 / N ) So we would like to see an estimate of N. But also what N.b. We need to use a “variable” here – N, about staff costs? What the expected number of customers. about infrastructure costs as we expand? Etc. etc. task 1b. Some updated information… We will charge £24/year for a subscription We will receive £1/month in ad revenue per user. It will cost £600k/ year for the site's data. Our researchers have also told us some infrastructure costs now: Hosting and management of the site will cost the company 50p/month for every subscriber using it. - Now how many customers do we need to break even? task 1b. Financial aggregation website Profit / customer in a year: £24 + £12 – (12 ✕ 0.50) = £30 So profit per customer/year is now: £30 – (£600,000 / N ) We break even when this is zero: 30 – (600,000 / N ) = 0 30 = 600,000 / N ⇒ N = 20,000 task 1b. Financial aggregation website So Profit / customer in a year: £24 + £12 - £6 = £30 And we calculated our “Break Even” user-base as: £600,000 / 30 = 20,000 - But how are we going to get those customers? Direct targeted marketing? Blanket marketing along with current mailshots? Advertising in relevant venues/publications, etc.? task 1c. Consider advertising to customers We can charge £24/year in subscription We get £1/month in ad revenue per customer. It will cost £600k/ year for the site's data. Site costs are 50p/month per customer. Direct mails cost 50p each. A 1% positive response rate is expected. - Now how many customers allow us to break even? - Does your analysis detect any problems here? (and if so any ideas for fixing these issues!) task 1c. Advertising to customers Our expected advertising cost / customer obtained is: £0.50 / 0.01 = £50 So our expected profit / customer is now: £30 - £50 = -£20 This is obviously somewhat of an issue being negative! - What can we do to make the venture profitable per customer? task 1c. Maximising profit / reducing costs Some options of what could be done about this…. Raise our subscription prices? Raise advertising prices? Reduce our infrastructure/bandwidth costs? Retain customers, rather than recapturing? other ideas here… task 1d. Customer retention Let's consider 2 options to make customers renew: a. At the end of the year send an email reminder about renewal (50% of subscribers will likely respond positively to this). b. Send an email offering an extra free year's subscription if they renew now (to which two thirds, or ~66%, of subscribers will respond positively). Both options cost 50p / customer to mail. Tip: consider users’ expected lifetime values… you may want to google -> Which is the better option for the venture, "geometric series" assuming lost customers will never return? task 1d. Customer retention option a a. Send an email reminder about renewing, to which 50% are expected to respond positively. We are multiplying the Remember Profit / customer in a year is: expected value of a customer £24 + £12 - £6 = £30 by 0.5 every year so we can use the formula… Expected lifetime revenue of a customer 1+ x + x2 + x3 + … = 1/(1-x) = £30 (1 + 0.5 + 0.25 + 0.125 + …) = £30 (1/(1-0.5)) = £60 task 1d. Customer retention option b b. Send an email offering an extra year's subscription if they renew now (to which two thirds, or ~66%, of subscribers will respond positively). Ok, so this means a person’s subscription will effectively be discounted to £12 / year, and the cohort's population only decreases every two years. Adjusted Profit / customer in an "option b" year is: £12 + £12 - £6 = £18 discounted subscription yearly ad revenue yearly hosting cost task 1d. Customer retention option b Note the cohort population only goes down every two years now… By altering the formula on the last slide we can get the new shortcut: task 1e. Customer lifetime value So option "b" seems preferable (£102 expected revenue per customer, versus £60). However, this assumes customers live forever. Imagine research shows that customers only ever renew subscriptions for a maximum of 10 years. Ok, so we know what the calculation should be, we just can't use the geometric series shortcut… task 1e. Customer retention over 10 years max task 1e. Customer lifetime value You will learn to automate these boring analytical equations, so that you can change parameters, and come up with new answers quickly! Let’s have a forward looking glimpse of how a scripting language might be used instead… (you can see the practical link on the Moodle page and try it yourself once we’ve learnt some scripting). task 1e. Customer lifetime value If we use python for this task results are in the order of the following: option a = £59.97 option b = £92.52 That's quite a big drop for option b, as it is a slower accruing series… but it is still higher so Option B is still preferred. task 1f. Moving on to analytics So this has illustrated some core business analysis – but how might modern analytics be able to improve the prospects of the venture? - What ways can you think up that could give us a more accurate analysis for the company? - CONSIDER THIS LIKE AN INTERVIEW QUESTION…. Advertising to real prospects? Our first option is better targeting of prospects! Recall that when trying to capture customers: Direct mails cost 50p each. A 1% response rate is expected. This rate is far too low - the company has extensive data about every potential customer’s transaction patterns. We should be applying: Customer Analytics and Market segmentation Targeted marketing Models Advertising to real prospects? Furthermore, N/LAB would have recommended a pilot mailing round, followed by: Basket Analysis and “modelling” of respondents to determine the features of those who subscribe A second round of mailing to customers we classify as "high-likelihood" subscribers based on the characteristics of those that renewed! Advertising to real prospects? Finally we suggest that the company try to find “influencers” and engage them via: Text Analytics Social Media Analytics -> What happens if with all of these modifications it is possible to improve initial uptake to 2%? Improving Renewal Rates! Recall the strategy: “Send an email reminder about renewing, to which 50% of subscribers are expected to respond to”. By this stage the company will have terabytes of data concerning the usage of all website subscribers: - Web Analytics Models and Behavioural Analytics Models - Followed by Targeted Marketing and “Sculpted Advertising” What happens if with all these modifications we can improve “option a” renewal rates up to 60%? Summary These are all types of modern business analytics. All these specializations have at their core the goal of making a prediction, based on previously seen data, to get insight, that will increase profits. These models have become essential to analysis and concentrate on CLASSIFYING a customer or FORECASTING how much they will spend: -> prediction + action The reading list Homework – Week 1 Read the Chapter supplied on Moodle before next week session: -> “Chapter 1: Data Analytic Thinking” This should take only 3 of the 5 hours of background study time this week, so keep reading or starting to script!