Data Science Strategy For Dummies

Ulrika Jägare is an M.Sc. Director at Ericsson AB. With a decade of experience in analytics and machine intelligence and 19 years in telecommunications, she has held leadership positions in R&D and product management. Ulrika was key to the Ericsson??s Machine Intelligence strategy and the recent Ericsson Operations Engine launch – a new data and AI driven operational model for Network Operations in telecommunications.

Foreword xv

Introduction 1

About This Book 2

Foolish Assumptions 3

How This Book is Organized 3

Icons Used In This Book 4

Beyond The Book 4

Where To Go From Here 5

Part 1: Optimizing Your Data Science Investment 7

Chapter 1: Framing Data Science Strategy 9

Establishing the Data Science Narrative 10

Capture 11

Maintain 12

Process 13

Analyze 14

Communicate 16

Actuate 17

Sorting Out the Concept of a Data-driven Organization 19

Approaching data-driven 20

Being data obsessed 21

Sorting Out the Concept of Machine Learning 22

Defining and Scoping a Data Science Strategy 26

Objectives 26

Approach 27

Choices 27

Data 27

Legal 28

Ethics 28

Competence 28

Infrastructure 29

Governance and security 29

Commercial/business models 30

Measurements 30

Chapter 2: Considering the Inherent Complexity in Data Science 31

Diagnosing Complexity in Data Science 32

Recognizing Complexity as a Potential 33

Enrolling in Data Science Pitfalls 101 34

Believing that all data is needed 34

Thinking that investing in a data lake will solve all your problems 35

Focusing on AI when analytics is enough 36

Believing in the 1-tool approach 37

Investing only in certain areas 37

Leveraging the infrastructure for reporting rather than exploration 38

Underestimating the need for skilled data scientists 39

`Navigating the Complexity 40

Chapter 3: Dealing with Difficult Challenges 41

Getting Data from There to Here 41

Handling dependencies on data owned by others 42

Managing data transfer and computation across-country borders 43

Managing Data Consistency Across the Data Science Environment 44

Securing Explainability in AI 45

Dealing with the Difference between Machine Learning and Traditional Software Programming 47

Managing the Rapid AI Technology Evolution and Lack of Standardization 50

Chapter 4: Managing Change in Data Science 51

Understanding Change Management in Data Science 52

Approaching Change in Data Science 53

Recognizing what to avoid when driving change in data science 56

Using Data Science Techniques to Drive Successful Change 59

Using digital engagement tools 59

Applying social media analytics to identify stakeholder sentiment 60

Capturing reference data in change projects 61

Using data to select people for change roles 61

Automating change metrics 62

Getting Started 62

Part 2: Making Strategic Choices for Your Data 65

Chapter 5: Understanding the Past, Present, and Future of Data 67

Sorting Out the Basics of Data 68

Explaining traditional data versus big data 69

Knowing the value of data 71

Exploring Current Trends in Data 73

Data monetization 73

Responsible AI 74

Cloud-based data architectures 75

Computation and intelligence in the edge 75

Digital twins 77

Blockchain 78

Conversational platforms 79

Elaborating on Some Future Scenarios 80

Standardization for data science productivity 80

From data monetization scenarios to a data economy 82

An explosion of human/machine hybrid systems 82

Quantum computing will solve the unsolvable problems 83

Chapter 6: Knowing Your Data 85

Selecting Your Data 85

Describing Data 87

Exploring Data 89

Assessing Data Quality 93

Improving Data Quality 95

Chapter 7: Considering the Ethical Aspects of Data Science 97

Explaining AI Ethics 98

Addressing trustworthy artificial intelligence 99

Introducing Ethics by Design 101

Chapter 8: Becoming Data-driven 103

Understanding Why Data-Driven is a Must 103

Transitioning to a Data-Driven Model 105

Securing management buy-in and assigning a chief data officer (CDO) 106

Identifying the key business value aligned with the business maturity 107

Developing a Data Strategy 108

Caring for your data 109

Democratizing the data 109

Driving data standardization 110

Structuring the data strategy 110

Establishing a Data-Driven Culture and Mindset 111

Chapter 9: Evolving from Data-driven to Machine-driven 113

Digitizing the Data 114

Applying a Data-driven Approach 115

Automating Workflows 116

Introducing AI/ML capabilities 116

Part 3: Building a Successful Data Science Organization 119

Chapter 10: Building Successful Data Science Teams 121

Starting with the Data Science Team Leader 121

Adopting different leadership approaches 122

Approaching data science leadership 124

Finding the right data science leader or manager 124

Defining the Prerequisites for a Successful Team 125

Developing a team structure 125

Establishing an infrastructure 126

Ensuring data availability 126

Insisting on interesting projects 127

Promoting continuous learning 127

Encouraging research studies 128

Building the Team 128

Developing smart hiring processes 129

Letting your teams evolve organically 130

Connecting the Team to the Business Purpose 131

Chapter 11: Approaching a Data Science Organizational Setup 133

Finding the Right Organizational Design 134

Designing the data science function 134

Evaluating the benefits of a center of excellence for data science 136

Identifying success factors for a data science center of excellence 137

Applying a Common Data Science Function 138

Selecting a location 138

Approaching ways of working 139

Managing expectations 141

Selecting an execution approach 142

Chapter 12: Positioning the Role of the Chief Data Officer (CDO) 145

Scoping the Role of the Chief Data Officer (CDO) 146

Explaining Why a Chief Data Officer is Needed 149

Establishing the CDO Role 150

The Future of the CDO Role 152

Chapter 13: Acquiring Resources and Competencies 155

Identifying the Roles in a Data Science Team 156

Data scientist 157

Data engineer 157

Machine learning engineer 158

Data architect 159

Business analyst 159

Software engineer 159

Domain expert 160

Seeing What Makes a Great Data Scientist 160

Structuring a Data Science Team 163

Hiring and evaluating the data science talent you need 165

Retaining Competence in Data Science 167

Understanding what makes a data scientist leave 169

Part 4: Investing in the Right Infrastructure 173

Chapter 14: Developing a Data Architecture 175

Defining What Makes Up a Data Architecture 176

Describing traditional architectural approaches 176

Elements of a data architecture 177

Exploring the Characteristics of a Modern Data Architecture 178

Explaining Data Architecture Layers 181

Listing the Essential Technologies for a Modern Data Architecture 184

NoSQL databases 184

Real-time streaming platforms 185

Docker and containers 185

Container repositories 186

Container orchestration 187

Microservices 187

Function as a service 188

Creating a Modern Data Architecture 189

Chapter 15: Focusing Data Governance on the Right Aspects 193

Sorting Out Data Governance 194

Data governance for defense or offense 195

Objectives for data governance 196

Explaining Why Data Governance is Needed 197

Data governance saves money 197

Bad data governance is dangerous 198

Good data governance provides clarity 198

Establishing Data Stewardship to Enforce Data Governance Rules 198

Implementing a Structured Approach to Data Governance 199

Chapter 16: Managing Models During Development and Production 203

Unfolding the Fundamentals of Model Management 203

Working with many models 204

Making the case for efficient model management 206

Implementing Model Management 207

Pinpointing implementation challenges 208

Managing model risk 210

Measuring the risk level 211

Identifying suitable control mechanisms 211

Chapter 17: Exploring the Importance of Open Source 213

Exploring the Role of Open Source 213

Understanding the importance of open source in smaller companies 214

Understanding the trend 215

Describing the Context of Data Science Programming Languages 215

Unfolding Open Source Frameworks for AI/ML Models 218

TensorFlow 219

Theano 219

Torch 219

Caffe and Caffe2 220

The Microsoft Cognitive Toolkit (previously known as Microsoft CNTK) 220

Keras 220

Scikit-learn 221

Spark MLlib 221

Azure ML Studio 221

Amazon Machine Learning 221

Choosing Open Source or Not? 222

Chapter 18: Realizing the Infrastructure 223

Approaching Infrastructure Realization 223

Listing Key Infrastructure Considerations for AI and ML Support 226

Location 226

Capacity 227

Data center setup 227

End-to-end management 227

Network infrastructure 228

Security and ethics 228

Advisory and supporting services 229

Ecosystem fit 229

Automating Workflows in Your Data Infrastructure 229

Enabling an Efficient Workspace for Data Engineers and Data Scientists 230

Part 5: Data as a Business 233

Chapter 19: Investing in Data as a Business 235

Exploring How to Monetize Data 236

Approaching data monetization is about treating data as an asset 237

Data monetization in a data economy 238

Looking to the Future of the Data Economy 240

Chapter 20: Using Data for Insights or Commercial Opportunities 243

Focusing Your Data Science Investment 243

Determining the Drivers for Internal Business Insights 244

Recognizing data science categories for practical implementation 245

Applying data-science-driven internal business insights 247

Using Data for Commercial Opportunities 248

Defining a data product 249

Distinguishing between categories of data products 250

Balancing Strategic Objectives 252

Chapter 21: Engaging Differently with Your Customers 255

Understanding Your Customers 255

Step 1: Engage your customers 256

Step 2: Identify what drives your customers 257

Step 3: Apply analytics and machine learning to customer actions 258

Step 4: Predict and prepare for the next step 259

Step 5: Imagine your customer’s future 260

Keeping Your Customers Happy 261

Serving Customers More Efficiently 263

Predicting demand 263

Automating tasks 264

Making company applications predictive 264

Chapter 22: Introducing Data-driven Business Models 265

Defining Business Models 265

Exploring Data-driven Business Models 267

Creating data-centric businesses 268

Investigating different types of data-driven business models 268

Using a Framework for Data-driven Business Models 275

Creating a data-driven business model using a framework 276

Key resources 277

Key activities 277

Offering/value proposition 278

Customer segment 278

Revenue model 279

Cost structure 280

Putting it all together 280

Chapter 23: Handling New Delivery Models 281

Defining Delivery Models for Data Products and Services 282

Understanding and Adapting to New Delivery Models 282

Introducing New Ways to Deliver Data Products 284

Self-service analytics environments as a delivery model 285

Applications, websites, and product/service interfaces as delivery models 287

Existing products and services 289

Downloadable files 290

APIs 290

Cloud services 291

Online market places 291

Downloadable licenses 292

Online services 293

Onsite services 293

Part 6: The Part of Tens 295

Chapter 24: Ten Reasons to Develop a Data Science Strategy 297

Expanding Your View on Data Science 297

Aligning the Company View 298

Creating a Solid Base for Execution 299

Realizing Priorities Early 299

Putting the Objective into Perspective 300

Creating an Excellent Base for Communication 300

Understanding Why Choices Matter 301

Identifying the Risks Early 301

Thoroughly Considering Your Data Need 302

Understanding the Change Impact 303

Chapter 25: Ten Mistakes to Avoid When Investing in Data Science 305

Don’t Tolerate Top Management’s Ignorance of Data Science 305

Don’t Believe That AI is Magic 306

Don’t Approach Data Science as a Race to the Death between Man and Machine 307

Don’t Underestimate the Potential of AI 308

Don’t Underestimate the Needed Data Science Skill Set 308

Don’t Think That a Dashboard is the End Objective 309

Don’t Forget about the Ethical Aspects of AI 310

Don’t Forget to Consider the Legal Rights to the Data 311

Don’t Ignore the Scale of Change Needed 312

Don’t Forget the Measurements Needed to Prove Value 313

Index 315