Index

[A][B][C][D][E][F][G][H][I][K][L][M][N][O][P][R][S][T][U][V][W][Z]

A

accelerometer data
Age of Implementation
AGI (Artificial General Intelligence), 2^nd, 3^rd, 4^th

AI (artificial intelligence)
  academic vs. industrial application of
  Age of Implementation
  AutoML
  business metrics
  capabilities of
    knowing results faster
    perceptual tasks
    prediction based on current trends
    structured data
    unstructured data
  causality vs. correlations, 2^nd
    complex systems
    counterfactuals
    incompleteness of data
    validity of causal relationships
  CLUE
  copying other organizations’ solutions, 2^nd
  data sources with differing levels of veracity
  evolution of definition
  failed projects
  fishing for something in data
  focus on infrastructure vs. on business actions
  high-level roles for
    as product
    automation of business process
    decision support system
    part of larger product
  human errors vs. AI mistakes
    actuarial vs. instance view
    domain actions
  human role in guiding to business results
  in physical systems
    IoT devices
    safety issues
    security issues
  knowledge base for project management
    AI skills vs. leadership skills
    prior technical knowledge and
  monetization with, 2^nd
    C part of CLUE
    general principles for
    limitations compared to humans
    medical diagnostics
    profit curve
    Sense/Analyze/React loop, 2nd
  new capabilities of
  nonprofit sector and
  purpose of book
  qualifications for leaders in
  relationship to machine learning, 2^nd
  relationship to other fields of business
  role of in business
  selecting and running projects
  Sense/Analyze/React loop
  starting with choice of analysis
  understanding possible business actions
  unicorns
    data engineering
    data science
    gap analysis
AI Effect
AI projects.
    See also machine learning (ML) pipelines.
  concluding
    three-state, yes/no/maybe classification of results
    timebox approach
  defining research question
    best practices for business leaders
    contractual language of technical domain
    understanding business concepts
  evolution of over time
    accounting for influence of time on profit curve
    including time directly in metrics
    timing diagrams
  failing fast
  impossible to stop in midair
  initiating AI efforts
    forces that teams are subject to
    look of failure
    look of success
    starting with simple projects
    starting with technically challenging projects
  pitfalls
    CLUE vs. gut feeling
    emulating large companies
    failing to build relationship with business team
    using advanced tools to look at big data
    using transplants
  prioritizing
    business metrics
    estimating project difficulty
    finding business questions for AI to answer
    methods and data
  software architecture
  traditional software systems vs.
    challenges amplified in AI projects
    challenges shared by
    machine learning pipeline in AI projects
    ossification of
AI Superpowers (Lee)
AI winters
air quality sensors
ambiguous (non-unique) profit curves
analysis roulette
anomaly detection
Architecture Tradeoff Analysis Model (ATAM)
ARIMA (autoregressive integrated moving average)
Artificial General Intelligence (AGI), 2^nd, 3^rd, 4^th
automation
  automated data analysis
    examples of
    poor business cases and
  automated ordering of supplies
  automating part of business process
  AutoML
  employment and
  non-monotonic profit curves
autonomous vehicles, 2^nd, 3^rd, 4^th
autoregressive integrated moving average (ARIMA)

B

Best Case/Worst Case analysis, 2^nd.
    See also MinMax analysis.
    See also MinMax analysis.
big data
  all necessary data vs.
  big data frameworks, 2^nd, 3^rd
  cleaning data
  data lakes
  using advanced tools to look at
bike rental system
  constructing profit curve for
  measuring progress on
black swans, 2^nd
Boeing
BPE (Business Performance Excellence) model
business actions.
    See also CLUE.
  best practices for business leaders
  considering why you haven’t taken
  fishing for something in data
  focus on infrastructure vs.
  linking AI capabilities with, 2^nd, 3^rd, 4^th, 5^th
  linking research questions with, 2^nd, 3^rd, 4^th
    business decisions based on technical metric
    questions answered by metric
    right business metric
    right research question
    understandability of business metric
  relationship with business team
  understanding and defining possible, 2^nd
  using to drive analysis
business metrics, 2^nd
  developing to summarize behavior of system
  examples of, 2^nd
  gut feelings vs., 2^nd
  impossible to stop projects in midair
  inability to define or select
  linking business and technology
  linking research questions with business problems
    business decisions based on technical metric
    determining you have the right business metric
    determining you have the right research question
    errors when defining business metrics
    G-MAFIA experiment
    mental calculations
    presenting technical metrics at business meetings
    questions answered by metric
    secondary impact of metrics
    surrogate metrics
    technical metrics that escape into the wild
    understandability of business metric
  linking technical progress with
    need for technical metrics
    profit curve
    understanding technical results in business terms
  measuring progress on AI projects
  measuring project success with
  organizational considerations when measuring technical progress
    arguments against profit curve
    information hoarding
    learning vs. being right
    profit curve improves over time
    profit curve precision depends on business problem
  random data vs.
  technical metrics vs.
  threshold
  time sensitivity
Business Performance Excellence (BPE) model
business process automation, 2^nd
  automating workflow
  job creation and
business questions.
    See also CLUE.
  balancing data, AI methods, and
  finding, 2^nd
  linking to research questions, 2^nd, 3^rd, 4^th

C

caret package
causality vs. correlations
  casual models vs. how AIs work
  complex systems
  counterfactuals
  incompleteness of data
  obstacles to wider use of
  validity of causal relationships
cleaning data
  big data
  improving algorithm vs., 2^nd
  machine learning pipeline, 2^nd, 3^rd, 4^th
cloud services, 2^nd, 3^rd, 4^th
CLUE, 2^nd
  Consider (available business actions)
    dependencies
    overview of
  Economize (resources), 2^nd, 3^rd
    dependencies
    overview of
  elements of, 2^nd
  gut feelings vs.
  infrastructure vs.
  Link (research question and business problem), 2^nd, 3^rd
    dependencies
    overview of
  MinMax analysis
    economizing resources devoted to
    general process for
  overview of, 2^nd
  size of company and
  substituting with different process
  Understand (the answer), 2^nd
    dependencies
    improving machine learning pipeline over time
    overview of
    understanding technical results in business terms
Computational Intelligence
cost plus pricing
counterfactuals

D

Dalio, Ray

data collection
  considerations for
  ideal data vs. collected data
  identifying collectable data
  in IoT setting vs. enterprise setting

data engineering and engineers
  cleaning data vs. improving algorithm
  estimating project difficulty
  ossification of machine learning pipeline
  safety issues
  team dynamics
  unicorns
data mining CRISP-DM process, 2^nd

data science and scientists
  acquiring skillsets of data scientists
  AI methods
  cleaning data vs. improving algorithm
  empathy for business audience learning technical metrics
  estimating project difficulty
  machine learning
  profit curve better defined by data science team
  relationship between technical and business metrics
  team dynamics
  unicorns
Data Science for Business (Provost and Fawcett), 2^nd
decision support system, AI as part of
  danger of
  helping management team
deep learning, 2^nd, 3^rd, 4^th, 5^th, 6^th, 7^th
  causality vs. correlations, 2^nd
  security issues
depreciation schedules
Derczynski, Leon
design of experiments (DOE)
Drucker, Peter

E

e-discovery
EDA (Exploratory Data Analysis)
ELT/ETL processes
expected value of perfect information, 2^nd
expert opinion estimates
Exploratory Data Analysis (EDA)

F

factory lines
  design of experiments
  machine learning pipeline for
failing fast, 2^nd
Fawcett, T.

G

gap analysis
Generative Adversarial Networks (GANs)
gesture recognition
global sensitivity analysis
  applicability of
  appropriateness of
  introducing errors
  not producing best possible result
GPS

gut feelings
  business metrics vs.
  CLUE vs.
  defined

H

Haenlein, M.
Hu, B.
Hubbard, D. W.
hypothesis testing

I

IBM Watson
image recognition, 2^nd, 3^rd, 4^th, 5^th, 6^th
information hoarding
Internet of Things (IoT) devices, 2^nd
  data collection

K

Kaplan, A.
Keras library
knowledge graphs

L

language translation
Lee, Kai-Fu
linear profit curves
local sensitivity analysis
  applicability of
  appropriateness of
  introducing errors
long short-term memory (LSTM)

M

machine learning (ML) pipelines, 2^nd, 3^rd
  AI projects vs. traditional software systems
    challenges amplified in AI projects
    challenges shared by
    ML pipeline in AI projects
    ossification of ML pipeline
  algorithm and data improvement
  balance vs. overfocusing
  complexity of real-world enterprise-strength pipelines
  evaluation metrics
  frameworks emerging from
  goal of
  how data scientists use
  lack of universal pipeline
  MinMax analysis
    determining you have the right ML pipeline
    economizing resources
    importance of
    interpreting results
    performing
    questions about
  need for analysis of
  ossification of
    addressing
    example of
  personalities take over in absence of data
  purpose of
  relationship to AI, 2^nd
  role of AI methods
  role of business leaders
  sensitivity analysis, 2^nd
    addressing interactions between pipeline stages
    CLUE
    common critiques to
    design of experiments
    detecting nonlinearity
    enhancing quality of data
    example of using results
    global sensitivity analysis
    local sensitivity analysis
    recent advancements in field of
  spanning whole communities
Maneuvering Characteristics Augmentation System (MCAS)
medical diagnostics, 2^nd, 3^rd
minimum viable product (MVP), 2^nd
MinMax analysis
  determining you have the right machine learning pipeline
  economizing resources devoted to
  effort to comprehend details of
  importance of
  interpreting results of
    decision to release product
    if machine learning pipeline needs improvement
    rules for
    smart parking meter example
  not limited to AI
  parts of
  performing
    best-so-far as upper limit
    estimates
    Max part
    Min part
    profit curves
    safety factors
  questions about
    can small companies/teams skipping MinMax
    use of term MinMax
    using MinMax as first analysis
    which part to perform first
  sensitivity analysis vs.
ML..
    See machine learning (ML) pipelines.
mlr package
Modified National Institute of Standards and Technology (MNIST)
monotonic profit curves, 2^nd
moonshots, 2^nd

N

natural language processing (NLP), 2^nd, 3^rd
  AutoML
  machine learning pipeline
Nest, 2^nd
Ng, Andrew
non-monotonic profit curves, 2^nd, 3^rd
non-unique (ambiguous) profit curves, 2^nd
nonlinearity

O

organizational silos
ossification of machine learning pipeline
  addressing
  causes of
  example of

P

pet monitors
pre-segmentation

product
  AI as fully autonomous product
  AI as part of larger, 2^nd
  evolution of capabilities of
  packaging AI as, 2^nd
  wide applicability of

profit curve
  accounting for influence of time on
  arguments against
  better defined by data science team
  constructing
  defined
  evolution of over time
  improvement over time
  in academia
  nonlinearity (convexity) in
  not limited to supervised learning
  precision depends on business problem
  sophistication of mathematical analysis

profit curves
  in MinMax analysis
    categories of
    categories of profit curves
    complex profit curves
    mental calculations
Provost, F.
pseudo experiments
publishing industry, 2^nd

R

recommendation engines, 2^nd, 3^rd
reinforcement learning

research questions
  defining
    best practices for business leaders
    contractual language of technical domain
    misaligned business and research questions
    understanding business concepts
  linking business problems and
    business decisions based on technical metric
    questions answered by metric
    right business metric
    right research question
    understandability of business metric
  linking business questions and, 2^nd
Roomba
root mean square error (RMSE) metric, 2^nd, 3^rd
rule engines, 2^nd, 3^rd

S

safety and security issues, 2^nd, 3^rd
  autonomous vehicles
  disagreement between AI engineering and safety engineering
  heuristics for building safe systems
  importance of human involvement
  in MinMax analysis
  local vs. global models
  non-monotonic profit curves
  ossification of machine learning pipeline
Sculley, D.
Sense/Analyze/React loop
  AI methods and data
  applicability of
  elements of
  finding business questions for AI to answer
  monetization with
  not limited to AI
  prioritizing projects
  speed in closing
sensitivity analysis, 2^nd
  CLUE
  common critiques to
  defined
  design of experiments
  detecting nonlinearity
  enhancing quality of data
    analyzing data-producing stage
    collapsing two stages of pipeline into one
  example of using results
  global sensitivity analysis
  increasing/decreasing accuracy by, 2^nd
  interactions between pipeline stages
    addressing
    effect of
  introducing errors
  local sensitivity analysis
  MinMax analysis vs.
  not limited to AI
  recent advancements in field of
  supervised vs.unsupervised learning
sentiment analysis
smart parking meters
smart speakers
smart thermostats, 2^nd, 3^rd
smart, internet-connected oven
speech recognition, 2^nd
stock market investments
streaming analytics
Support-Vector Machines (SVMs), 2^nd

T

team dynamics
technical metrics
  business metrics vs.
  escaping into the wild
  linking to business metrics, 2^nd
  need for
  poorly understood
  presenting directly
technology smokescreens
Tesla
three-state, yes/no/maybe classification of results
threshold
timebox approach
timing diagrams
transplant projects
trend estimates

U

uncanny valley concept
unicorns
  acquiring skillsets of
  data engineers
  data science
  gap analysis

V

vacuum cleaning robots, 2^nd

value threshold
  defined
  evolution of over time, 2^nd
  improving machine learning pipeline over time
  MinMax analysis
  selecting, 2^nd, 3^rd
  sensitivity analysis, 2^nd
  timing diagrams
video surveillance systems, 2^nd
  ossification of machine learning pipeline
voice recognition, 2^nd

W

Wikimedia Foundation

Z

zero-shot learning