MIS EXAM 2 STUDY GUIDE
- IN("column names") : - IN(number, number) :
- checks for set/list inclusion (identical) - only returns records that match those 2 numbers
How would you evaluate GE's new software platform Predix? What are the key challenges of building such a platform?
Challenges include connecting all HW devices (many different segments), a lot of data, dealing with different customers, business data privacy concerns, and replacing talent. Predix did not work and was spun off.
Comparison operators:
=, >=, <. !=
6. Be able to decide the correct service model (among IaaS, PaaS, and SaaS) if given a scenario.
>
What is foreign key? When given a set of tables and certain assumptions, understand how to identify foreign key.
A foreign key is an attribute or a set of attributes in a relation of a database that serves as the primary key of another relation in the same database and helps link different relations in the same database.
When given a real-world problem, be able to propose various data mining techniques to solve the problem.
,
When given a traditional industry, be able to discuss how to make digital transformation for companies in the industry, as well as the potential challenges associated with such digital transformation.
,
For Amazon, Microsoft, and Google, what are the potential synergies of cloud business with their non-cloud businesses? MSFT
- All MS Office product features (most successful products) are dependent on the Cloud: people store their files on the Cloud, Bing & LI give MSFT more data and more control over the ecosystem - The enterprise software application is the biggest synergy for Office - Bing search engine has AI capabilities and companies can leverage the language model - Gaming is data intensive and similar to streaming
For Amazon, Microsoft, and Google, what are the potential synergies of cloud business with their non-cloud businesses? Amazon
- Cloud stores 3rd party seller data and user data on AWS, which leads to better recommendations and customer experience - AWS does data intensive jobs (ML, computerization) - Supply chain optimization with delivery network
For Amazon, Microsoft, and Google, what are the potential synergies of cloud business with their non-cloud businesses? GCP
- Least diverse portfolio, as 80% of their revenue comes from ads - Google stores user data in the Cloud and pushes their own ads based on that data - Google is good at ML/AI and is trying to build combine capability between ads/AI
Understand different ways to measure the distance between two clusters.
- Most popular distance measure = Euclidean distance[A1] - Minimum distance (single linkage) = distance between the pair of records that are closest - Maximum distance (complete linkage) = distance between the pair of records that are farthest - Average distance (average linkage) = average distance of all possible distances between records in one cluster and records in the other cluster - Centroid distance = distance between 2 cluster centroids o Centroid = vector of measurement averages across all the records in that cluster
What are the benefits faced by IT vendors in offering cloud computing?
-Instantly deploy bug fixes and product enhancements to all users (centralized management of resources) -Accessibility to anyone with an internet connection =larger customer base -For SaaS, reduced costs associated with development across different platforms, software piracy, and distribution
Be able to list example applications of digital twins in certain industries or scenarios.
-Manufacturing: BMW + NVIDIA: allows engineers and data scientists to collaborate in real time but digitally -Autonomous vehicles: can use digital twins to train autonomous vehicles (high stakes environment) -Supply chain -Healthcare
What are the challenges faced by IT vendors in offering cloud computing?
-Software development costs (learning, search for new info) -Form a new sales force -Pay as you go model means revenue is spread out over time (while for packaged software revenue is front-loaded) -Subscription based model -Users get flexibility, which is a challenge for the provider in terms of server capacity investment
When given a specific scenario, be able to discuss key challenges of implementing Industry 4.0 in such a scenario?
-Technology -capability -vendor selection -scalability -cybersecurity -Platform and standards -legacy systems -cloud -data models -integration -regulatory approval -Customers -agility -expectations -blue oceans -customizing -Competition -first adopter -first follower -laggard -Benefits -labor productivity -inventory -capital investment -operating costs -intangible benefits -ROI -Management -vision -leadership -governance -functional/technical/collaboration -resources -Implementation -migration -contingency planning -Network buy in (upstream/downstream) -company wide rollout People -resistance to change -skepticism -training/expertise -winners/losers -recruiting
Be able to interpret classification tree when given one example.
.
Why would a firm offer a loyalty card to its customers? Why do we say online systems would give organizations more insights than traditional transaction processing systems?
A firm (like Costco) would offer a loyalty card to link transactions to customers. A cash register alone cannot track who buys what. Online systems give organizations more insights than traditional transaction processing systems because purchasing and browsing behavior are tracked. This tracking leads to increased data that can give insights into consumer behavior and the company can change its strategies (product mix, webpage layout, recs, etc) to increase sales.
What is primary key? What is the difference between simple primary key and composite primary key? When given a table and certain assumptions, understand how to identify primary key.
A primary key is an attribute or set of attributes that uniquely identifies A simple primary key is comprised of a single attribute (i.e. Customer ID) while a composite primary key is comprised of multiple attributes (i.e. Customer ID + Product ID + Date).
How do AI technologies and humans complement each other in organizational decision making?
AI gives insights from historical data, while humans make decisions from an intuitive approach. In terms of uncertainty, humans make intuitive decisions in the face of unknowns, while AI provide access to real time information. In terms of complexity, humans decide where to seek and gather data and choose among options with equal data support, while AI collects, curates, processes, and analyzes data. In terms of equivocality, humans negotiate, build consensus, and rally support, while AI analyze sentiments and represent diverse interpretations. AI: Prediction, Correlational, Analytical, Productive, Inductive Human: Judgement, Causal, Intuitive, Creative, Deductive
Boolean operators:
AND, OR, NOT o Order of evaluation: NOT first, then AND, then OR o Use parenthesis to change order of operations (parenthesis get evaluated first)
What is Initial Coin Offering (ICO)? How does it differ from Initial Public Offering (IPO)?
An ICO is how new projects can raise funds by selling their underlying crypto tokens in exchange for fiat currencies, Bitcoin and/or Ether. ICOs are more cost effective than an IPO. In an IPO, you lose flexibility and control over your company to equity investors.
What is association detection? What are the key metrics to look at when you decide the best association rule?
Association detection is an unsupervised learning data mining technique that draws trends/patterns from a data set and can determine which behaviors or outcomes go together in market basket analysis or affinity analysis. The antecedent is something that has already happened; the consequent is what you are working on or trying to promote. Key metrics are the lift ratio, confidence, and support.
What is blockchain and what are its key characteristics?
Blockchain is a new platform technology to verify and record transactions among an interconnected set of users. Each transaction between parties is a "block" and the cumulative set of transactions across the entire network is the "chain." The key characteristics are: - All data are not changeable, meaning all transactions records are transparent and traceable, which enables the creation of a secure, private, and tamper-resistance single source of truth - No centrally coordinating entity - All transactions are broadcast across the network to be verified and validated
What is business analytics? Why has business intelligence become so popular nowadays?
Business analytics is using data to predict future trends based on data visualization, statistical modeling, and sophisticated data analysis for organizational decision making. The purpose is to understand the past and present to predict the future. Business analytics has become popular due to the increasing connectivity of devices/systems, the declining cost of storage, declining cost of computing, and the increased availability of data. This increased data can yield results that can be used for competitive advantage. [Observe (data collection) à Analyze (statistical modeling) à Action (is it meaningful?)]
What are business processes? What is process mining? How does process mining help improve an organization's workflow and business processes?
Business processes are a sequence of tasks or activities that produce desired outcomes, such as procurement (buy) à production (make) à fulfillment (sell). Business processes are processes interrelated with other processes. Process mining is using a data-driven approach to discover, validate, and improve workflow. Ex: Celonis process mining = software company that provides process mining for Dell and other companies. 4 major activities of process mining = collection of time-stamped log data, discovery of actual processes, enhancement of existing processes to improve business outcomes, and monitoring of changes for additional growth opportunities. The use of data helps identify opportunities for efficiency and to realize those efficient changes. Data flow helps discover non-obvious patterns from data.
Be able to list key areas where businesses can apply various data mining techniques.
Businesses use data mining in marketing and promotion targeting (which customers will respond to an offer), market basket analysis (which products customers buy together), collaborative filtering (how to make recommendations and personalize individual customer experience based on trends exhibited by similar customers), and hiring and promotion.
What is NPV? Understand how to calculate NPV given an IT investment scenario.
NPV is Net present value (NPV) is the difference between the present value of cash inflows and the present value of cash outflows over a period of time. NPV is the result of calculations that find the current value of a future stream of payments, using the proper discount rate. In general, projects with a positive NPV are worth undertaking while those with a negative NPV are not.
How do contract service agreements (CSAs) differ from outcomes-based service models?
CSAs: -Deliverable: Preventive maintenance of a GE asset ("break fix" insurance) to keep asset running optimally -Deal: More standardized -Terms: Contract fee; Length = life of asset -Margins: high -Access to customer operating data: Some -Knowledge of customer business model: Some Outcomes-based: -Deliverable: Optimize assets and customer's processes -Deal: idiosyncratic -Terms: revenue/profit sharing -GE's margins: high if outcomes reached -Access to customer operating data: significant -Knowledge of customer's business model: significant
What is classification? Understand how to interpret a classification tree, i.e., be able to describe how you would make predictions when given a classification tree.
Classification is a data mining technique under supervised learning. It is a special form of regression where you arrange data into predefined groups. Y is not a continuous variable but a categorical variable. It shows how likely a variable will fall into (A, B, or C) category.
What is clustering? What is the difference between classification and clustering?
Clustering is an unsupervised learning data mining technique where similar records are grouped tougher based on certain characteristics. For example, market segmentation is identifying customers with similar buying behavior. Classification: We use Y Clustering: We don't know Y and use distance metrics to put records into groups
What are the advantages provided by ICO to raise funds?
Its advantages are that funds can be raised at the early stage of projects, entrepreneurs have full ownership and control over the company, and they are more cost effective than an IPO. In an IPO, you lose flexibility and control over your company to equity investors. An ICO gives more flexibility in terms of raising additional funds by issuing more digital tokens.
What is data mining?
Data mining is the automated search in large databases for non-obvious patterns and relationships to anticipate events or predict outcomes. We cut large data sets into different parts (trimming, validation, and testing) in order to focus on the best prediction power of the model. Businesses use data mining in marketing and promotion targeting (which customers will respond to an offer), market basket analysis (which products customers buy together), collaborative filtering (how to make recommendations and personalize individual customer experience based on trends exhibited by similar customers), and hiring and promotion.
What is database? Why do people use a database?
Database is a single table or collection of related tables. People use a database to retrieve information from multiple tables and eliminate data redundancy. Database fits into an organization in the form of informational reports (data about different function areas) that are analyzed and used to find trends, which are used to make business decisions.
How does predictive analytics differ from descriptive analytics? How does prescriptive analytics differ from predictive analytics?
Descriptive -What happened? -Understand past patterns -Reports -Chart, pivot table Predictive -What will happen? -How do you act on the correlation? -Machine learning/data mining Prescriptive -How do you make it happen? -Causal inference & reinforcement learning -Ex: Recommendation system = optimizing reward
What are digital twins? How are digital twins related to but different from simulations?
Digital twins are a true-to-reality virtual representation of a real-world physical asset or system, which is continuously updated. They are used for predictions and to optimize with the virtual representation of physical objects. Digital twins have an updated real time connection and are more complex than a 1 scenario simulation. Physical twins are when you attach sensors and chips to get data and connect to digital twins.
What is an efficient supply chain and what is a responsive supply chain? Which one does Zara focus on and why?
Efficient: -Predict demand and supply demand efficiently at lowest cost -Maximize capacity utilization (so excess inventory is possible, then need to minimize inventory) -Suppliers selected based on cost and quality -Push model as manufacturing strategy Responsive: -Respond quickly to demand changes (so insufficient inventory is possible, then need to minimize stock out) -Minimize lead time aggressively -Supplier selected based on speed, flexibility, and quality -Pull model as manufacturing strategy Zara focuses on a responsive supply chain to minimize lead time and quickly respond to consumer preferences and keep up with the rapidly changing fashion industry.
What is ERP? What is CRM? What is SCM? Understand what each of these enterprise systems basically does.
Enterprise applications address the needs of multiple users throughout an organization or work group. The ERP, CRM, AND SCM are part of the enterprise applications that are connected to an organization's database to ensure consistent data formats across the enterprise. The Enterprise Resource Planning (ERP) system integrates many functions needed in an organization. It faces internally and requires customization to each industry to support business processes. The Supply Chain Management software solves uncertainties with supply chain with regards to communication, reduce costs, reduce waiting, reduce uncertainty, and better metrics. It faces partners. The Customer Relationship Management (CRM) software supports customer-related sales and marketing activities and faces partners.
When given a specific industry/scenario, be able to discuss the implications of AI such as generative models for jobs in that industry/scenario.
Examples in class: using ChatGPT to create a job post (HR), prepare a resume, ask for a salary raise. Jobs like copywriters, computer programmers, HR, sales, and lawyers will all be affected by AI.
When given certain features of a product, be able to match it with the right supply chain strategy.
Fashion: -Hedonic, expressive taste-making -Constantly changing what is "in" and "out" -Diffuses quickly through adoption curve -Developed and seeded by extreme consumers -Usually responsive/pull Basics: -Utilitarian, functional taste-following -Stable over time, what is "in" today is "in" tomorrow -Mass market -Need based -Usually efficient/push
What is IaaS? For this service model, what are the resources the cloud vendor will provide/manage and what are the resources the cloud user will have a control? What basic question is it essentially addressing for a cloud user?
For Infrastructure as a Service, the cloud vendor will provide a server (raw processing power and/or storage). The cloud user will have control over the OS and the applications, but the IaaS provider controls the hardware. The question it answers is the customer's need for a server. The top 3 players in IaaS are AWS, Microsoft Azure, and Google Cloud Platform. This requires a huge upfront investment because you need almost an unlimited server capacity, making it harder for small firms to enter.
What is PaaS? For this service model, what are the resources the cloud vendor will provide/manage and what are the resources the cloud user will have a control? What basic question is it essentially addressing for a cloud user?
For Platform as a Service, the cloud vendor will provide a platform for app developers to run their apps. The cloud user will have control over the applications, but the PaaS provider controls the HW and the OS. The question it answers is the customer's need for a place to run their apps. The top 3 players in IaaS are AWS, Microsoft Azure, and Google Cloud Platform. This requires a huge upfront investment because you need almost an unlimited server capacity, making it harder for small firms to enter.
What is SaaS? For this service model, what are the resources the cloud vendor will provide/manage? What basic question is it essentially addressing for a cloud user? Know that the cloud user will usually not manage or control the cloud infrastructure, operating systems, and even individual application capabilities, except for limited user-specific configuration settings.
For Software as a Service, the cloud vendor provides a place to run software. The customer wants to use all 3 layers of the ecosystem, which is usually provided by a group of companies because it is a very complex ecosystem for 1 company to provide on its own. The question it answers is the customer wanting to access a software app via their thin client interface (e.g. web browser). Younger and smaller firms like Slack and Zoom provide services in SaaS.
What are the key elements of GE's Industrial Internet? How does the Internet of Things interact with analytics to create value?
GE's Industrial Internet is adding sensors to its machines, connecting to a common cloud-based platform, and investing in modern software/analytics capabilities. The 3 key players are intelligent machines, advanced analytics, and people at work. Instrumented industrial machine à extraction and storage of proprietary machine data à industrial data systems à machine-based algorithms and learning analysis à big data analytics à remote and centralized data visualization à data sharing with the right people and machines à physical and human network à intelligence flows back into machines.
What is hierarchical clustering? What are the pros and cons of hierarchical clustering when compared with k-Means?
Hierarchical clustering is when you start with each cluster comprising exactly 1 record and then progressively combine the 2 nearest clusters until there is just 1 cluster left at the end, which consists of all the records.
Understand why merely using big data and predictive analytics may not be enough for product design in the fashion industry.
Human insights helps spot trends, make trends, and have an intuition of consumer preference. Big data and predictive analysis cannot predict the consumer demand in the future for this industry because it is cyclical and rapidly changing. The data driven approach is based on 2 assumptions: consumer preferences are stable over time & consumer preferences are generated by consumers themselves.
What are the sources of AI bias?
Minority bias is when we have limited data on a certain minority group that leads to a non-representative training sample and thus non-representative model that may leave minority groups out of hiring/recruiting, for example, or fail to make accurate healthcare diagnoses. Missing data bias is when data is missing in a non-random fashion. An example is when customer spending is a function of income and some social classes could choose not to report income. Obtaining data can be difficult in many domains, such as diagnosis, hiring decisions, and legal cases. Another source of bias is a feedback loop, If users accept biased results from the model, when the next time the model is trained, it will learn to continue these mistakes.
What are the effects of cloud computing on innovation and entrepreneurship?
IaaS/PaaS providers enable digital entrepreneurship (to SaaS providers) and help level the playing field by allowing organizations of all sizes to embrace digital transformation and new business models to compete. Organizations no longer need to pay huge infrastructure costs for enterprise systems.
What is industry 4.0? What are the digital technologies that could be included in Industry 4.0?
Industry 4.0 is the convergence and application of various digital technologies. The digital technologies that can be included in Industry 4.0 are: advanced robotics, additive manufacturing, augmented reality, simulation, horizontal/vertical integration, industrial internet, cloud, cybersecurity, and big data and analytics.
What is the Internet of Things? Understand that IoT allows for digital transformation of previously analog machine and service operations.
IoT allows for digital transformation of previously analog machine and service operations. This can be done by adding sensors to collect data that can be stored on the Cloud and then used for analytics. The Internet of things describes physical objects with sensors, processing ability, software and other technologies that connect and exchange data with other devices and systems over the Internet or other communications networks.
For association detection, what does lift ratio indicate? What does a lift ratio less than 1 imply? What does confidence indicate? What does support count indicate? Suppose you plan to launch a cross-selling campaign and are given a set of selling strategies (deriving from association rules), be able to identify and argue which strategy you would go with.
Lift ratio is an indicator of whether an association rule is useful. It is found by comparing the confidence with the probability of only the consequent item. Lift ratio = Prob (consequent|antecedent)/Prob(consequent). When the lift ratio is >1, we do the promotion. A lift ratio less than 1 indicates that doing that promotion would decrease sales. A lift ratio > 1 indicates that a promotion will help lift sales. A lift ratio = 1 indicates independence. Confidence indicates at what rate consequents will be found given that an antecedent has occurred. It is Prob (consequent|antecedent). It can tell you how likely a cross-selling strategy will be successful. Support count indicates the frequency of items being purchased in the database, or an item's popularity.
What is omnichannel?
Omnichannel is the integration of online and offline interactions with customers.
Should GE make its new software platform open or closed? What are the arguments for either choice?
Open: -If a platform is too closed, it is hard to connect with other businesses -Open platforms encourage adoption and connectivity Closed: -Data is sensitive -Consumers don't want to give up privacy -Better performance due to control
What are opportunities and challenges associated with building logistics blockchain for Tetra Pak and its partners in the supply chain network?
Opportunities: Connected solutions offers customers more granularity in product traceability, tracking a single product through the supply chain to identify every step on the way. Blockchain will ensure safety and transparency between all players. Challenges: Companies are hesitant to try new approaches without evidence of success. Also, the risk and volatility associated with blockchain technology, as well as general lack of knowledge surrounding it pushes companies away from adoption. Additionally, lack of integration with existing systems and talent to manage it adds another layer of risk.
Suppose now your company wants to adopt an enterprise software application (e.g., CRM, ERP) and you have two options - buying some packaged enterprise software or using software-as-a-service model (i.e. getting on-demand enterprise software). What are the benefits of buying the packaged one and what are the benefits of getting the on-demand one? Understand how the size of your organization affects your choice.
Packaged enterprise software Benefits: -customization -single access point = more secure -full control Software as a service model (on-demand) Benefits: -Low IT cost and maintenance (launches upgrades seamlessly) -flexibility/access -Higher operating income -IaaS/PaaS provides cost effective place to host apps -Only pay for what you need How the size of the organization affects the choice: Large organizations benefit from packaged enterprise software because they get full integration, customization, and they can afford it. Smaller companies can benefit from SaaS due to the flexibility of the pay-as-you-go model and it does not require the upfront huge cost of an enterprise software.
What are potential data sources for an organization?
Potential data source include a transaction processing system (i.e. Costco memberships, Amazon purchasing and browsing behavior are tracked). Tracking leads to increased insights and increased sales. Companies can use data to change their product mix, webpage layout, recommendations, etc. Other data sources include enterprise software (CRM, SCM, ERP), process mining (using IT to track the entire business process and identifying bottlenecks), external sources (social media, partners, data aggregators like Acxiom - collects real estate, census reports, credit card applications), data sharing with retailers that have information about the customer (manufacturers don't), and cross application data.
Understand the basic idea of k-Means algorithm.
Pre-specify a number of clusters k and assign each case to one of k clusters so as to minimize a measure of dispersion within the clusters. 1. Start with k clusters 2. At every step, each record is re-assigned to the cluster with the closest centroid 3. Recompute the centroids of clusters that lost or gained a record, and repeat step 2 4. Stop when moving more records between clusters increases cluster dispersion.
Predictive vs. prescriptive analytics
Predictive: -Regression -Data science + stat = predictive -Linear function Prescriptive: -Casual testing -Data science + machine learning = prescriptive -A/B testing = changing the value of X and seeing how that leads to different Ys -Randomized experiments that require time and money
What are the key challenges of outcomes-based services?
Some questions related to outcomes-based services are: Difficulty measuring results (what is the counterfactual?), difficulty sharing the client's proprietary financial and operational data, new customer relationship management, new expertise and sales talent, heterogeneity across customers in different sectors, and taking on consumer risk over the lifetime of their purchases.
Recall your experience in running the CloudStrat simulation. Understand how to be consistent across your product orientation, competitor orientation, and market orientation.
Product: focus on a specific product (Legacy/SaaS) or diversifying - Invest in SaaS performance to bring SaaS to feature-parity of above relative to legacy offerings Competitor: focus on winning market share or maximizing profit - Grow the ecosystem so developers prefer LegaSoft or Netware Market: focus on enterprise, mid-market, or small business segments - Lower prices attract mid market and small businesses - Enterprise = maintain prices and increase margins
What are the pros and cons of data-driven approach for product design in the fashion industry?
Pros: -Reflects market preferences -Eliminates guesswork -Helps more efficiently manage inventory and production Cons: -Can this approach predict the consumer's future demand in the industry? -Data can only reveal what worked, NOT what will work; cannot be used to set trends and tastes -Bias in the algorithm?
What are the major benefits and drawbacks for such a supply chain structure?
Pros: Cheap labor costs from outsourcing Cons: -Lower quality -Cannot reach high end market -Need to respond to preferences very quickly -Long lead time -May have to sell extra inventory at a discount = lower profit margin -Inventory kills -Large quantity
Pros and Cons of Hierarchical Clustering
Pros: Flexible in # of clusters (not pre-specified), which helps be adaptable in the context of preference/constraints Cons: Requires calculation of all possible combinations of distance, which consumes a lot of computing power and is subject to outlier problems
Pros and cons of k means
Pros: More popular because more efficient and requires less computing power. Cons: If you want to change the number of clusters, you have to re-cluster
What is relational database? Understand the meaning of the following terms in relational database context: records, attributes, fields. What are the requirements for a table in relational database?
Relational database are when tables are related based on common keys. The rows in a relational database are called records. The named columns are called attributes or fields. The requirements for a table in relational database are: - Table must have a unique name - Every attribute fall into a known domain or legal values (i.e. age should be a number) - Attributes in tables must have unique names - Every record must be unique (no 2 rows with the same values for all fields)
When given a specific industry/scenario, be able to discuss the implications of AI such as generative models for jobs in that industry/scenario - HR
Roles taken by AI -automating routine tasks (performance reporting, Q&As with potential new recruits, creating training manuals) -content/sentiment analytics to understand worker concerns, comments, and new ideas Roles taken by human -recruitment, active listening, coaching, advising, communication, employee relations, performance management
When given a specific industry/scenario, be able to discuss the implications of AI such as generative models for jobs in that industry/scenario - Computer programmers
Roles taken by AI -checking bugs and errors, creating short routines, breaking down bigger coding tasks into small components Roles taken by human -determine software design, new features, business problems that can be solved by software, particularly in the area of enterprise software, which is highly complex and requires deep understanding on the business processes and user needs
When given a specific industry/scenario, be able to discuss the implications of AI such as generative models for jobs in that industry/scenario - Lawyers
Roles taken by AI -creating summaries of case notes/relevant laws, drafting documents such as contracts or agreements Roles taken by human -face to face communication and interviews with clients, negotiation, developing legal strategies
When given a specific industry/scenario, be able to discuss the implications of AI such as generative models for jobs in that industry/scenario - copyrighters
Roles taken by AI -identifying useful elements, relevant content, and good examples Roles taken by human -conveying brand voice, meaning, and personality in a way that is original, authentic, and human
When given a specific industry/scenario, be able to discuss the implications of AI such as generative models for jobs in that industry/scenario - Sales
Roles taken by AI -info entry into CRM, handle basic/standard inquiries, prepare for sales docs, client data analytics, etc Roles taken by human -relationship building, identifying business needs, developing salesmanship techniques and strategies
Know the order of SELECT, FROM, WHERE in SQL statement
SELECT [DISTINCT] column_list FROM table_list [WHERE conditional expression]
How does the big data approach influence the three Gap's brands differently?
Since Old Navy supplies more commodity clothing, they are trend followers and can use more of a data driven approach because it is utilitarian, stable, and need-based. GAP is in the middle in terms of price and quality. Banana Republic is a branded good price and quality, meaning it is a trend setter or taste maker and the big data approach is not as effective because this side of the market is constantly changing what is in and out.
What are smart contracts? Understand how smart contracts can be used in logistics and supply chain areas.
Smart contracts are event-triggered, automated pieces of computer code intended to facilitate, verify, or enforce the negotiation of a contract. They are more complex transactions and are often permissioned or closed for more control over the complexity of transactions. They are advantageous in that they are self-executing (more efficient due to no human labor/error) and because everything is transparent, all parties must comply. Smart contracts are beneficial in logistics and supply chain because they can ensure the compliance with a contract by 2+ parties because the terms of the contract are event-triggered, which incentivizes both parties to act accordingly. An example of smart contracts in logistics and supply chain is Walmart Canada and 300cubits in the shipping industry. 300cubits uses smart contracts in a Booking Deposit Module to match shippers to shipping companies and then issue a booking reference. After this, cargo delivery status is received and analyzed and digital tokens are distributed based on the smart contract to ensure compliance.
What are smart, connected products? What benefits can smart and connected products bring to a focal company and its customers?
Smart, connected products are physical components (mechanical and electrical parts), "smart" components (sensors, microprocessors, data storage, controls, software, embedded OS and user interface), and connectivity components (ports, antennae, and protocols enabling wired or wireless connections with the product. The capabilities of smart, connected products are monitoring, control, optimization, and autonomy. - Monitoring = sensors and external data enable monitoring of product condition, external environment, and product's operation and usage, as well as alerts/notifications of changes - Control = software embedded in the product or Cloud that enables control of product functions and personalization of user experience - Optimization = monitoring and control enable algorithms that optimize product operation in order to enhance product performance, allow predictive diagnostics, service, and repair - Autonomy = monitoring + control + optimization allows autonomous product operations, self-coordination with other systems, autonomous enhancement and personalization, and self-diagnosis and service
What are the differences between statistics, machine learning, and artificial intelligence (AI)?
Stats: Builds model from data insights. Simplest model, with biggest error. The problem with complex models is that they are not efficient, hard to interpret, and can overfit new data. ML: The goal is the best prediction power. ML automatically searches for the best model. It has good predicting power (training data) and good performance (test data). Compared to AI, this is a more primitive machine that gives prediction results. (data + insights) AI Uses insights to make a decision. This is a broader concept and it is faster. Questions: Who takes responsibility? (data + insights + making decisions)
What is SQL?
Structured Query Language (SQL) is an international standard language for processing a database, used by all major data base management systems. To retrieve information from the database, run a query.
What are Tetra Pak's digital efforts?
Tetra Pak's digital efforts are the smart factory, connected workforce, and connected solutions. The smart factory is used to discover, interpret, and communicate meaningful patterns in data of the end-to-end supply chain using sensor data, digital twins for optimization, and preventative maintenance. They connect manufacturing equipment to IOT, run reports on the equipment, and test solutions on digital twins. The connected workforce provides information at the fingertips of our workers via mobile and wearable technology (can use VR/AR to access digital tools). Connected solutions enhance or create new customer solutions by adding digital capabilities to the products and services. For example, turn the package into an information channel to enhance manufacturers' marketing intelligence as well as the consumer experience (uniquely identifiable, more direct communication channel with the consumer, gathering deeper actionable insights into consumer behavior).
What is cloud computing? What are the essential characteristics of cloud computing?
The Cloud is a set of resource infrastructures accessible over the internet. Cloud computing is a model for enabling on-demand network access to a shared pool of configurable computing resources (i.e. servers, storage, applications) that can be rapidly provisioned with minimal service provider interaction. Cloud is more economical than ERP because you don't own anything; you only need to subscribe and can quickly access your data. When software applications are not within their own data centers, companies rent data and raw processing power. They develop on top of cloud companies (SaaS running on top of cloud infrastructure) and the user can use a mobile device to access.
What are the right metrics to value cloud companies?
The key problem is how to account for revenue generated through Cloud software contracts and how to assess how much future revenue investors can rely on. New metrics are annual recurring revenue, level of churn, and net dollar retention (how much a company generates in a given year from customers it had the year before).
What basic functions should an ERP system provide to an educational institution? What are the major problems in ERP implementation in educational industry in general and what are the unique problems for the San Diego City Schools (SDCS) in particular?
The ERP should provide student performance assessment, reporting applications, financial applications, project management, procurement and delivery, HR management applications, food and health services applications, and SIS. All of these would be connected to a central database. The major problems in ERP implementation in educational industry are that even the best SIS modules only have 50% of required functionality and in this industry there is an average failure rate of 505 upon ERP implementation. IT vendors make more money selling to for-profit organizations with deep pockets, so they are not interested in serving non-profit organizations. Because users cannot afford the high cost of ERP implementation, these initiatives fail and thus there is high risk associated with it. The key problems with ERP and SDCS is that there are limited budgets, low IT spending, under qualified personnel, and changing leadership.
What is Metaverse? How are digital twins related to but different from Metaverse?
The Metaverse is a collective virtual shared space connected to many aspects of the physical world, including people, places, and things, thereby enabling shared experiences across physical and digital worlds. IT is seen in gaming, entertainment/streaming, shopping etc. Sone other elements of metaverse are digital currency, NFTs, online shopping workplace, etc. Both digital twins and the metaverse meet real-time connection between the digital and physical world, but the metaverse is more oriented toward consumers. In the metaverse you can create things that are not necessarily the same as the physical world. Digital twins are more valuable if they are more accurate to reality/need to mimic real world as much as possible. Digital twins are related to metaverse due to the necessary human component of both concepts Digital twins is the industrialization of complex product development, manufacturing, processes etc + humans. Metaverse is game, entertainment, shopping, etc. + humans. Humans create Cloud, IOT, VR/AR, AI/ML platforms.
What are the barriers to cloud computing adoption?
The barriers to cloud computing are: -security (concerns about data confidentiality on the Cloud) -availability (dependence on Internet connection, long term viability of partner firms, and risk of infrastructure crashing) -total cost of ownership ($ per CPU per hour, # of accesses, # of data transfers) -lock in leads to high switching costs (large amounts of data are hard to move) -interoperability with your own systems -less flexibility in choosing different versions (possible training costs and shifts in operating procedures) and customization compared to packaged software.
What are the benefits to users from cloud computing?
The benefits to users of adopting cloud computing are increased efficiency because it reduces a firm's assets that can stay idle and depreciate over time (reduced capital investment), it reduces IT maintenance and upgrade costs, and frees the firm's IT resources for core competency tasks. Other benefits are rapid access to resources/rapid development on platforms and scalability, the ability to adjust resource consumption on a per-need basis.
What are the challenges and risks of using AI?
The challenges and risk of AI is the 80/20 rule, in that 80% of predictions are done well, 20% are unusual events, and we don't want to risk being wrong in high stakes situations (i.e. autonomous driving, healthcare). Other challenges are the demand for enormous amounts of computing power (traditional CPUs vs. graphics processing units vs. quantum computing), legal issues for generative models (pictures/copyrights), AI bias, AI interpretability/explainability, and the responsibility and trust issue.
What is the conventional supply chain structure in the fashion industry? What are the major benefits and drawbacks for such a supply chain structure?
The conventional supply chain structure in the fashion industry is: suppliers - manufacturers - distributors - retailers - customers.
When given a specific scenario/industry, be able to discuss the key challenges of implementing blockchain applications in such a scenario/industry?
The key challenges for Blockchain applications are technical, market/business, behavioral/educational, and legal/regulatory. - Technical: o Underdeveloped ecosystem infrastructure o Lack of mature applications o Scarcity in developers o Immature middleware and tools o Scalability o Legacy system o Lack of standards - Market/business o Moving assets to blockchain o Critical mass of users o Quality of startups o Volatility of crypto o Not enough qualified individuals o Cost issues - Behavioral/educational o Lack of understanding of potential value o Limited executive vision o Change management o Trusting a network o Few best practices - Legal/regulatory o Unclear regulations o Government interference o Compliance requirements
What are the main challenges faced by 300cubits for its blockchain applications?
The main challenges faced by 300cubits for its blockchain applications are: - New system requires a lot of coordination of different stakeholders and challenges on their existing systems/business processes, leading to slow adoption - 2 sided platform subject to network effects - No existing laws and standards to regulate the implementation of smart contracts - No wide adoption due to talent shortage, risk (volatile crypto market), compatibility, legal, and costly B2B transactions.
Based on the SDCS case, what are the major benefits from the ERP investment? Among these benefits, what are the intangible benefits from the ERP investment? Understand the importance of incorporating intangible benefits (including productivity increase and soft benefits) into investment on return analysis.
The major benefits from the ERP investment are yearly cost savings (salary savings, overtime cost savings, print shop related cost savings, reduction in processing errors), productivity improvement in clerical staff and principals, and soft benefits such as improved employee morale/productivity (less turnover), improved recruiting (hire potentially high-quality candidates), and improved management access to data. The intangible benefits are the productivity increases and soft benefits of improved employee morale/productivity (less turnover), improved recruiting (hire potentially high-quality candidates), and improved management access to data.
What are the major human resource (HR) functions incorporated in an ERP system?
The major functions are benefits (medical, dental, and life insurance; retirement benefits), payroll (time and labor collecting, paycheck processing), and personnel (applicant tracking system, position control, review certification and demographics).
What are the potential participants on 300cuibits' blockchain platform? How would you help 300cuibits to attract more participants to the platform?
The potential participants are shippers, shipping companies, cryptocurrency investors, and equity investors. 300cubits should address the needs of different stakeholders at the appropriate time and promote TEU token accordingly. Shippers could use TEU tokens to make shipping bookings. Shipping companies can use TEU tokens to confirm the shipment booking. Cryptocurrency investors could buy digital tokens as an investment. Equity investors may be interested in the equity of the company. 300cubits can attend industry events to promote TEU tokens and gain access to key players. They can educate shippers and shipping companies on how the system can help solve their problems. They can alternate between appealing to the shipping industry and the crypto investors. If the crypto market is going up, more resources should be invested in addressing crypto investors. If the crypto market is going down, more resources should be invested in the industry players who will actually use the digital tokens.
Why did SDCS choose to implement a "human resource management" module as a starting point?
They did not start with SIS because this was a high stake/mission critical project with a high failure rate. They wanted to start with a lower stake module (HRM). If they succeeded, then they could replicate it. The HRM would provide productivity improvement and soft benefits. IT was also the most compatible with their already existing systems.
What is the major problem faced by the container shipping industry? How can blockchain be used to address this problem?
This industry lacks collateral for booking deposits, which caused the long-lasting shipper no-show and cargo roll-over problems. They couldn't require deposits due to the very competitive nature of the market, where competitors will undercut and not charge a deposit, meaning shippers will migrate to those companies. Blockchain can be used to address this problem because an Ethereum-based cryptocurrency, TEU, can serve as collateral. Using smart contracts that distribute digital tokens, companies are forced to comply with event-triggered contracts with the collateral incentive in order to fulfill the transaction.
What is the rationale for Zara to offer most products in limited run?
To cultivate exclusivity, encourage customers to buy right away and at full price, encourage customers to visit more often, and reduce risk of mistakes.
What are the four major data mining techniques? If you are given an example, be able to know which type of data mining technique it belongs to.
Unsupervised learning: draw inferences from datasets where there is no-predefined outcome variable 1. Association detection: Market basket analysis/affinity analysis: determine which behaviors/outcomes go together 2. Clustering: Similar records are grouped together based on certain characteristics a. Ex: market segmentation = identify customers with similar buying behavior b. Hierarchical clustering = Start with each cluster comprising exactly 1 record and then progressively combine the 2 nearest clusters until there is just 1 cluster left at the end, which consists of all the records [A1] c. K-means = Pre-specify a desired number of clusters, k, and to assign each case to one of k clusters so as to minimize a measure of dispersion within the clusters Supervised learning: draw inferences from datasets in which an outcome variable of interest is known (Y) and the algorithm is to learn how to predict this value with new records where the outcome variable is unknown 3. Regression: Attempt to find a function which estimates the relationships among variables with the least error 4. Classification: Arrange data into predefined groups; y is a categorical variable (not continuous) a. Classification Trees
What are the major upfront and ongoing costs of implementing HR modules?
Upfront = software, infrastructure, labor (training, consulting). Ongoing = on site IT support, software maintenance, hardware maintenance, incremental networking.
Why did Walmart want to adopt blockchain technology for its logistics?
Walmart wanted to adopt blockchain for its logistics to address the vast data discrepancies in the invoice and payment process for freight carriers, for which they used some of its own trucking fleet as well as 3rd party carriers. More than 70% of their invoices required reconciliation efforts. The use of multiple information systems between Walmart Canada and its carriers meant there was no integration and it was labor intensive. With blockchain, they could capture information at every step from the tender offer from the carrier to proof of delivery and the approval of payment, providing a single source of truth for shippers and carriers. This led to a less than 1% invoice dispute. Walmart Canada was able to involve key stakeholders, coordinate among multiple parties, and integrate the blockchain with other technologies such as IoT for operational improvements.
For supervised learning, why do we need to partition data into training data, validation data, and testing data?
We are focused on creating a model with the best prediction power. The training data builds the model. The algorithm needs to learn to predict the Y value where the outcome variable is unknown. The validation data evaluates and adjusts the models. The testing data re-evalutes the models. The validation and testing data test if the function can predict the Ys. The new data where X is known but Y is not is fed into the model to get Y, which is what you use to predict/classify using the final model.
When you decide if you want to get service from a particular cloud vendor, why should you consider the long-term viability of the cloud vendor?
When you decide if you want to get service from a particular cloud vendor, you should consider the long-term viability of the cloud vendor because you want to make sure they have sufficient server capacity investment in order to accommodate for flexibility required from the user. You want to make sure there is not a risk of the crash of their infrastructure and a potential loss of your data.
What is the difference between Zara's supply chain and the conventional supply chain in the fashion industry?
Zara produces 40% of the fabric (raw material) in house and dyes (raw material) are purchased from its own subsidiary (suppliers). 60% of Zara's merchandise produced is in-house OR outsourced to contract manufacturers (manufacturers). Zara controls the distributor and retailer parts. This leads to a much shorter lead time for Zara and limited production of most products.
Why does Zara want to build an integrated offline and online store model?
Zara wants an integrated store model to give a seamless customer experience, aggregate demand to manage uncertainty on demand allocation, enable inventory pooling and reduce inventory costs, create flexibility, reduce logistics and reverse logistics costs, and increase data and experiments.
How does Zara's online store change its supply chain structure?
Zara's online store led to the addition of an Integrated DC, which takes in inventory from Global DCs as well as overflow from Flagship Stores. In the Integrated DC, it has an online DC that is used to ship to online orders and an overflow warehouse that ships inventory to flagships stores for pick up or re-stocking. Before this, Zara's supply chain included a Global DC that sent inventory directly to flagship stores.