Literature & Lectures

Question 1

What is a HiPPO?

Answer 1

the highest-paid person’s opinion.

Question 2

What is the idea of ‘Mute the HiPPO’?

Answer 2

Businesses should rely more on data instead of HiPPOs

Question 3

The 5 management challenges by McAfee?

Answer 3

1. Leadership: Establish leadership teams that set clear goals, define what success looks like, and ask the right questions.
2. Talent Management: As data become cheaper, the complements to data become more valuable
3. Technology: The tools available to handle the volume, velocity and variety of big data.
4. Decision-Making: Put information and the relevant decision rights in the same location.
5. Company culture: Move away from acting solely on hunches and instinct and breaking bad habits

Question 4

Three analytics capability levels proposed by LaValle

Answer 4

1. Aspirational: Use analytics to justify actions
2. Experienced: Use analytics to guide actions
3. Transformed: Use analytics to prescribe actions

Question 5

5 recommendations of LaValle

Answer 5

1. Focus on the biggest and highest-value opportunities
2. Within each opportunity, start with questions, not data
3. Embed insights to drive actions and deliver value
4. Keep existing capabilities while adding new ones
5. Use an information agenda to plan for the future

Question 6

What does Hadoop do?

Answer 6

combines commodity hardware with open-source software. It takes incoming streams of data and distributes them onto cheap disks; it also provides tools for analyzing the data.

Question 7

What is a relational database?

Answer 7

a database structured to recognize relations among stored items of information.

Question 8

What is PoS Data?

Answer 8

Point of Sale Data is data collected by a business when a transaction happens.

Question 9

What is preventive maintenance?

Answer 9

uses sensor data to monitor a system, then continuously evaluates it against historical trends to predict failure before it occurs.

Question 10

3 Business Models enabled with Big Data

Answer 10

1. Differentiating creates new experiences
2. Brokering augments the value of information
3. Delivery networks enable the monetization of data

Question 11

The differences between hierarchical and relational databases as pointed out by Lake & Crowther (2013)?

Answer 11

In a hierarchical database, segments are implicitly joined with each other.

In a relational database, this relationship between tables is captured by foreign keys and primary keys.

Question 12

a hierarchical database

Answer 12

segments are implicitly joined with each other

Question 13

a relational database

Answer 13

this relationship between tables is captured by foreign keys and primary keys.

Question 14

SQL database

Answer 14

relational database (handle structured data)

Question 15

NoSQL database

Answer 15

non-relational database (handle unstructured data)

Question 16

4 Pros of SQL database

Answer 16

1. Fits structured data readily into well- organized tables
2. Are more mature & represent huge investments by vendors and users
3. Convenient for transactions that require great precision (supports ACID*)
4. Offers a big feature set and data integrity

Question 17

4 Cons of SQL database

Answer 17

1. Scaling. They aren’t designed to function with data partitioning
2. Complexity. When data doesn’t fit easily into a table, the database’s structure can be complex, difficult and slow to work with.
3. Can entail large amounts of complex code and doesn’t work well with modern, agile development.
4. Large feature set. Users don’t need all the features, as well as the cost and complexity they add.

Question 18

What does ACID stands for?

Answer 18

Atomicity means an update is performed completely or not at all

Consistency means no part of a transaction will be allowed to break a database’s rules

Isolation means each application runs transactions independently of other applications

Durability means that completed transactions will persist

Question 19

4 Pros of NoSQL database

Answer 19

1. Efficient handling of unstructured data
2. Easier to work with for developers that are not familiar with SQL
3. Enable better performance
4. Process data faster than relational databases

Question 20

4 Cons of NoSQL database

Answer 20

1. May cause problems for applications that require great precision (ACID)
2. Require manual query programming (is fast for simple tasks but is difficult and time- consuming for others)
3. Don’t offer the degree of reliability because it doesn’t support ACID
4. May compromise consistency because it doesn’t support ACID

Question 21

What is SQL?

Answer 21

Is Structured Query Language, which is a computer language for storing, manipulating and retrieving data stored in a relational database.

Question 22

Why would you choose NoSQL?

Answer 22

To enable cloud, easily scalable, real-time, with shorter processing time and relatively cheaper than RDBMSs

Question 23

Why Hadoop?

Answer 23

(1) It is scalable: you can add more nodes on the fly

(2) It is fault-tolerant: if nodes go down, data gets processed by another node

Question 24

What are the 3 core components of Hadoop?

Answer 24

1. Hadoop Distributed File System (HDFS)
   The storage layer of Hadoop
2. Map-Reduce
   The data processing layer of Hadoop. It processes data in two phases (1) Map Phase and (2) Reduce Phase
3. Yet Another Resource Locator (YARN)
   The resource management layer of Hadoop

Question 25

Name 4 reasons why Hadoop is important

Answer 25

1. Hadoop has a better Career Scope 2. A Maturing Technology 3. Data managing 4. Omnipresent 5. Open for all (Hadoop is easy to manage) 6. Professionals shortage

Question 26

What is a Database Management System? (DMS)

Answer 26

A software system that enables users to define, create, maintain and control access to databases.

Question 27

2 Logical structures for database models

Answer 27

1. Navigational models (e.g. hierarchical-, network-, and graph database models) 2. Relational models

Question 28

What is understood when we speak about a table?

Answer 28

a collection of records

Question 29

What does cardinality tell us?

Answer 29

Describes the expected number of related occurrences between the two entities in a relationship

Question 30

What is an Entity-Relationship database? (ER)

Answer 30

Schematic representation of the database

Question 31

What is a normalisation process in programming?

Answer 31

Fitting messy ‘real-life’ data into a homogenized and uniformized database. Making sure that the database is accurate, scalable, easy to update and queried.

Question 32

What does a normalisation process prevent?

Answer 32

redundancy, confusion, improper keys, wasted storage, incorrect/outdated data

Question 33

5 stages of the Big Data Life Cycle (Most companies focus only on the modeling/analysis part but focusing on all processes will result in higher business rewards)

Answer 33

1. Data acquisition: produce, derive and collect data 2. Information extraction and cleaning: Pull out information and express it in a structured form suitable for analysis 3. Data integration, aggregation and representation: Collection of heterogeneous data from multiple sources 4. Modeling and analysis: Methods for querying and mining big data 5. Interpretation: A decision-maker had to interpret the results of the analysis

Question 34

The inherent challenges of Big Data

Answer 34

1. Heterogeneity 2. Inconsistency and incompleteness 3. Scale 4. Timeliness 5. Privacy and data ownership

Question 35

7 fundamental concepts of Big Data by Provost & Fawcett

Answer 35

1. Extracting useful knowledge from data by following a process with reasonably well-defined stages. The Cross-Industry Standard Process for Data Mining (CRISP-DM) 2. Evaluating data-science results requires careful consideration of the context in which they will be used. 3. The relationship between the business problem and the analytics solution often can be decomposed into tractable subproblems via the framework of analyzing the expected value 4. Information technology can be used to find informative data items from within a large body of data 5. Entities that are similar with respect to known features or attributes often are similar with respect to unknown features or attributes 6. If you look too hard at a set of data, you will find something—but it might not generalize beyond the data you’re observing ''overfitting'' 7. To draw causal conclusions, one must pay very close attention to the presence of confounding factors, possibly unseen ones

Question 36

4 mistakes most managers make with analytics by Lambrecht & Tucker

Answer 36

1. Not understanding the issues of integration 2. Not realizing the limits of unstructured data 3. Assuming correlations mean something 4. Underestimating the labor skills needed

Question 37

Two key advices of Frank

Answer 37

1. Start small: define a few relatively simple analytics, this allows the organization to see what the data can do. Also, the results are easier to test. 2. Targeted prototyping Capture only the data you need to perform the test, instead of dealing with all of the data available. This is a lower-risk way to see what big data can do for your firm and to test your firm’s readiness to use it.

Question 38

The top benefits of AI

Answer 38

Make better decisions Optimize internal operations Optimize external operations Free workers to be more creative Enhance current products

Question 39

4 Factors that drove the AI wave

Answer 39

1. New algorithms for machine learning 2. The internet and the cloud 3. Big data 4. Moore’s Law

Question 40

What is Neutral Language Processing? (NLP)

Answer 40

NLP is a term for everything from speech recognition to language generation, each requiring different techniques (such as chatbots and translations)

Question 41

5 fundamental methods and techniques for AI

Answer 41

1. Heuristics Heuristics are a way to employ a practical method to find a solution that is not guaranteed to be optimal, but one that is sufficient for the immediate goals (such as navigation) 2. Support Vector Machine Classification problems where there is no straight rule for identifying the classes. (such as a spam filter or identifying handwriting or characters) 3. Artificial Neutral Networks Understanding complex relationships between features of a certain item (such as image or speech recognition) 4. Markov Decision Process Find a policy for the decision-maker, tell him which particular action should be taken at which state. Solving complex decision-making problems (such as inventory planning) 5. Neutral Language Processing NLP is a term for everything from speech recognition to language generation, each requiring different techniques (such as chatbots and translations)

Question 42

What is Cognitive Computing used for?

Answer 42

The ability to understand

Question 43

What does a Linear Regression do?

Answer 43

Linear Regression allows us to map numeric inputs to numeric outputs, fitting a line into the data points.

Question 44

What is the essence of an algorithm?

Answer 44

To capture the dominant trend and fit our line within that trend. Note: We always want to find the trend, not fit the line to all the data points!!!

Question 45

What does a bias-variance trade-off mean?

Answer 45

Machine Learning models fulfill their purpose when they generalize well. Generalization is bound by the two undesirable outcomes — high bias and high variance. A situation with low bias and low variance represents the desired situation ''generalization'' This trade-off is the most integral aspect of Machine Learning model training. Detecting whether the model suffers from either one is the sole responsibility of the model developer.

Question 46

Explain underfitting

Answer 46

!! High Bias !! Underfitting is the case where the model has “ not learned enough” from the training data, resulting in low generalization and unreliable predictions.

Question 47

Explain overfitting

Answer 47

!! High Variance !! Overfitting is the case where the overall cost is really small, but the generalization of the model is unreliable. This is due to the model learning “too much” from the training data set.