Contents

Section Title Page
1 Overview 6
1.1 Citation 6
1.2 Have Questions? 6
2 Why Driverless AI? 7
3 Key Features 8
4 Supported Algorithms 10
5 Installing and Upgrading Driverless AI 12
6 Launching Driverless AI 13
6.1 Messages 14
7 The Datasets Page 14
7.1 Adding Datasets 15
7.2 Dataset Details 17
7.2.1 Dataset Details Page 17
7.2.2 Dataset Rows Page 19
7.2.3 Modify by Recipe 19
7.2.4 Downloading Datasets 20
7.3 Splitting Datasets 21
7.4 Visualizing Datasets 22
7.4.1 The Visualization Page 22
8 Running an Experiment 26
8.1 Before You Begin 26
8.2 New Experiment 26
8.3 Completed Experiment 30
8.4 Model Scores 31
8.4.1 Experiment Summary 33
8.5 Viewing Experiments 36
8.5.1 Checkpointing, Rerunning, and Retraining 36
8.5.2 Deleting Experiments 39
9 Diagnosing a Model 3
10 Project Workspace 41
10.1 Linking Datasets 42
10.1.1 Selecting Datasets 42
10.2 Linking Experiments 43
10.2.1 New Experiments 43
10.2.2 Checkpointing Experiments 44
10.3 The Experiments Leaderboard 44
10.3.1 Leaderboard Scoring 45
10.3.2 Comparing Experiments 46
10.4 Unlinking Data on a Projects Page 48
10.5 Deleting Projects 48
11 Interpreting a Model 48
11.1 Interpret this Model button – Regular Experiments 49
11.2 Interpret this Model button – Time-Series Experiments 50
11.2.1 Multi-Group Time Series MLI 50
11.2.2 Single Time Series MLI 52
11.3 Model Interpretation – Driverless AI Models 54
11.4 Model Interpretation – External Models 57
11.5 Understanding the Model Interpretation Page 59
11.5.1 Summary Page 61
11.5.2 DAI Model Dropdown 61
11.5.3 Random Forest Dropdown 77
11.5.4 Dashboard Page 80
11.6 General Considerations 81
11.6.1 Machine Learning and Approximate Explanations 81
11.6.2 The Multiplicity of Good Models in Machine Learning 82
11.6.3 Expectations for Consistency Between Explanatory Techniques 82
12 Viewing Explanations 83
13 Score on Another Dataset 86
14 Transform Another Dataset 86
15 The Driverless AI Scoring Pipelines 88
15.1 Visualize the Scoring Pipeline 88
15.2 Which Pipeline Should I Use? 90
15.3 Driverless AI Standalone Python Scoring Pipeline 91
15.3.1 Python Scoring Pipeline Files 91
15.3.2 Quick Start – Recommended Method 92
15.3.3 Quick Start – Alternative Method 93
15.3.4 The Python Scoring Module 96
15.3.5 The Scoring Service 96
15.3.6 Python Scoring Pipeline FAQ 99
15.3.7 Troubleshooting Python Environment Issues 99
15.4 Driverless AI MLI Standalone Scoring Package 100
15.4.1 MLI Python Scoring Package Files 101
15.4.2 Quick Start – Recommended Method 102
15.4.3 Quick Start – Alternative Method 102
15.4.4 Prerequisites 102
15.4.5 MLI Python Scoring Module 104
15.4.6 K-LIME vs Shapley Reason Codes 105
15.4.7 MLI Scoring Service Overview 105
15.5 Driverless AI MOJO Scoring Pipeline 108
15.5.1 Prerequisites 108
15.5.2 MOJO Scoring Pipeline Files 109
15.5.3 Quickstart 109
15.5.4 Execute the MOJO from Java 110
15.5.5 MOJO Scoring Pipeline – C++ Solution 112
15.5.5.1 Downloading the Scoring Pipeline Runtimes 112
16 Deployment 115
16.1 Additional Resources 116
16.2 Deployments Overview Page 116
16.3 AWS Lambda Deployment 116
16.3.1 Driverless AI Prerequisites 116
16.3.2 AWS Access Permissions Prerequisites 116
16.3.3 Deploying the Lambda 118
16.3.4 Testing the Lambda Deployment 119
16.3.5 AWS Deployment Issues 120
16.4 REST Server Deployment 121
16.4.1 Prerequisites 121
16.4.2 Deploying on REST Server 121
16.4.3 Testing the REST Server Deployment 123
16.4.4 REST Server Deployment Issues 124
17 About Driverless AI Transformations 125
17.1 Numeric Transformers 125
17.2 Time Series Experiments Transformers 126
17.3 Categorical Transformers (String) 127
17.4 Text Transformers (String) 128
17.5 Time Transformers (Date, Time) 129
18 Logs 129
18.1 Sending Logs to H2O 133
19 References 133
20 Authors 135

Start Your 21-Day Free Trial Today

Get It Now
Desktop img