- A
- accessing
- Amazon S3 using AWS CLI, 199–200
- AWS, 154–160
- objects, 191–195
- Account menu (AWS management console), 158–160
AccountOverview
intent, customizing, 308–312
acl
object, 183
ACMEBankBotAccountOverview
function, 286–299
ACMEBankBotTransactionList
function, 299–304
- activity detection, in Amazon Rekognition, 422
- agent, 11
- algorithms
- Amazon CloudFront, 153–154
- Amazon Cognito
- about, 142, 201
- common tasks, 204–217
- identity pools, 203, 218–219
- key concepts, 201–203
- retrieving client secret, 213–214
- user pools, 202–213, 218–219
- Amazon Cognito Federated Identities, 203
- Amazon Comprehend
- about, 140–141, 257
- interactive text analysis using AWS CLI, 262–265
- key concepts, 257–260
- text analysis using Management Console, 260–262
- using with AWS Lambda, 266–274
- Amazon DynamoDB
- about, 142, 221
- adding items to tables, 228–231
- common tasks, 225–236
- creating an index, 231–233
- creating tables, 225–228, 278–285, 433–435
- as event source for AWS Lambda, 240
- key concepts, 221–225
- performing queries in, 235–236
- performing scans in, 233–235
- Amazon Glacier, 183
- Amazon Lex
- about, 141, 275
- creating bots, 278–315
- key concepts, 275–278
- Amazon Machine Learning
- about, 141, 317
- creating datasources, 321–332
- creating ML models, 337–341
- creating models with, 319
- creating real-time prediction endpoints, 346–347
- key concepts, 317–321
- making batch predictions, 341–346
- making predictions using AWS CLI, 347–349
- using real-time prediction endpoints, 349–350
- viewing data insights, 332–337
- Amazon Polly, 141
- Amazon Rekognition
- about, 141, 421
- analyzing images using Management Console, 423–428
- interactive image analysis with AWS CLI, 428–433
- key concepts, 421–423
- using with AWS Lambda, 433–444
- Amazon S3 (Simple Storage Service)
- about, 142, 181
- accessing using AWS CLI, 199–200
- bucket versioning, 197–199
- common tasks, 185–200
- as event source for AWS Lambda, 240
- key concepts, 181–185
- Amazon SageMaker
- about, 141–142, 353
- creating notebook instances, 357–362
- deploying DNN classifiers using Google TensorFlow Estimators API and, 408–418
- key concepts, 353–357
- notebook instances, 357–362
- preparing test and training data, 362–364
- SDK for Python, 354
- training DNN classifiers using Google TensorFlow Estimators API and, 408–418
- training models using built-in algorithms on dedicated training instances, 379–384
- training Scikit-learn models on dedicated training instances, 368–378
- training Scikit-learn models on notebook instances, 364–368
- using Google TensorFlow with, 387–418
- Amazon Transcribe, 141
- Amazon Translate, 141
- Amazon Web Services (AWS)
- about, 135
- cloud computing, 135–136
- cloud deployment models, 138–139
- cloud service models, 136–138
- ecosystem of, 139–142
- free-tier accounts, 142–148
- global infrastructure, 151–160
- management console, 156–160
- Anaconda Navigator, 4
- analyzing
- images using Amazon Rekognition Management Console, 423–428
- images with AWS CLI, 428–433
- ANN (artificial neural network), 388–389
- API sets, in Amazon Rekognition, 422
- application virtualization, 136
- artificial neural network (ANN), 388–389
- ASR (Automatic Speech Recognition), 141
- attributes
- about, 318
- in Amazon DynamoDB, 222
-
auc()
function, 126
- authentication, 201
- authorization, 201
- Autoencoding, 10
- Automatic Speech Recognition (ASR), 141
- availability
- of Amazon Comprehend, 259–260
- of Amazon Lex, 278
- of Amazon Machine Learning, 321
- of Amazon Rekognition, 423
- of Amazon SageMaker, 357
- as an advantage of cloud computing, 136
- of AWS Lambda and, 244
- availability zones (AZ), 151–153
- AWS. See Amazon Web Services (AWS)
- AWS boto3 SDK, 354
- AWS CLI
- accessing Amazon S3 using, 199–200
- Amazon SageMaker and, 354
- detecting entities with, 263–264
- detecting key phrases with, 264–265
- interactive image analysis with, 428–433
- making predictions using, 347–349
- sentiment analysis with, 265
- text analysis with, 262–265
- AWS DeepLens, 142
- AWS IAM, 142
- AWS Lambda
- about, 142, 237
- common tasks, 244–254
- common use cases for, 237–238
- creating functions, 285–304, 435–444
- creating Python functions using AWS Management Console, 244–250
- deleting Python Lambda functions using AWS Management Console, 253–254
- key concepts, 238–244
- testing Python functions using AWS Management Console, 250–253
- using Amazon Comprehend with, 266–274
- using Amazon Rekognition with, 433–444
- AWS Management Console
- access to, 168
- creating Python Lambda functions using, 244–250
- deleting Python Lambda functions using, 253–254
- testing Python Lambda functions using, 250–253
- AWS public datasets, 30
- AWS re:Invent, 139
- AWS Service, 173
- AWS Signature V4, 417–418
-
axes
class, 69
- axes object, 55–56
-
Axes.pie()
method, 70
- axis labels, 56
- axis object, 56
- AZ (availability zones), 151–153
- B
- bar charts
- about, 61–62
- grouped, 62–65
- stacked, 65–67
- stacked percentage, 67–68
- batch learning, 11–12
- batch predictions, making, 341–346
- batch transform, in Amazon SageMaker, 355
- behavioral biometrics, machine learning for, 7
- Bezos, Jeff, 139
- binary classification models, 119–126, 318
- Black, Benjamin, 139
- Boston house prices dataset, 29
- bots
- about, 275
- creating in Amazon Lex, 278–315
- testing, 314–315
- box plots, 71–73
-
boxplot()
function, 39–40, 72
- BPaaS (business process as a service), 137–138
- Breast cancer dataset, 30
- broad network access, as a characteristic of cloud computing, 135
- bucket versioning, in Amazon S3, 197–199
- buckets
- about, 181–182
- creating, 185–189
- built-in algorithms
- Amazon SageMaker, 356
- training models on dedicated training instances using, 379–384
- built-in slot types, 277
- business process as a service (BPaaS), 137–138
- C
- changing
- permissions associated with existing groups, 172–173
- storage classes of objects, 195–196
- chatbots. See bots
- choosing
- algorithms, 23–24
- hyperparameter values, 131–132
- input features, 18–22
- churn prediction, machine learning for, 7
- claims, 202
- classification, 6
- classification models
- binary, 119–126
- evaluating, 119–131
- multi-class, 126–131
- client
- in authentication flow, 202
- secret, retrieving, 213–214
- client application, 276
- cloud computing, 3, 135–136
- cloud deployment models, 138–139
- cloud service models, 136–138
- clustering, 9
- community cloud, 138
- computation graph, 387
- compute requirements, AWS Lambda and, 239
-
confusion_matrix()
function, 123, 128
- consistent reads, in Amazon DynamoDB, 224
- constant leaf nodes, 388
- context, AWS Lambda and, 242
- continuous tasks, reinforcement learning for, 11
-
corr()
function, 40
- costs
- Amazon S3, 183
- as an advantage of cloud computing, 136
- creating
- AWS Lambda functions, 244–250, 285–304, 435–444
- batch predictions in Amazon Machine Learning, 341–346
- bots in Amazon Lex, 278–315
- buckets, 185–189
- datasources in Amazon Machine Learning, 321–332
- identity pools (Amazon Cognito), 214–217
- indexes in Amazon DynamoDB, 231–233
- linear regression models with Google TensorFlow, 390–408
- machine learning models with linear regression, 86–92
- machine learning models with logistic regression, 101–109
- machine learning models with Scikit-learn, 79–113
- ML models in Amazon Machine Learning, 337–341
- models with Amazon Machine Learning, 319
- new features, 44–46
- notebook instances (Amazon SageMaker), 357–362
- predictions using AWS CLI in Amazon Machine Learning, 347–349
- real-time prediction endpoints in Amazon Machine Learning, 346–347
- roles, 173–176
- tables in Amazon DynamoDB, 225–228, 278–285, 433–435
- test datasets, 80–86
- test datasources, 330–332
- training datasets, 80–86
- training datasources, 324–330
- user pools (Amazon Cognito), 204–213
- users, 167–172
- credit scoring, machine learning for, 7
- Cristianini, Nello (author)
- An Introduction to Support Vector Machines and Other Kernel-based Learning Methods, 96
- cross-validation, 22
- custom slot types, 277
- customizing
-
AccountOverview
intent, 308–312
-
ViewTransactionList
intent, 312–314
- D
- data acquisition, 5
-
data
attribute, 27
- data channels (Amazon SageMaker), 355
- data formats (Amazon SageMaker), 356
- data insights, viewing, 332–337
- data munging, 18
- data preprocessing techniques
- about, 31
- creating new features, 44–46
- handling missing values, 42–44
- obtaining data overview, 31–41
- one-hot encoding categorical features, 47–50
- transforming numeric features, 46–47
- Data Transfer Pricing (Amazon S3), 183
- data visualization
- about, 51
- common plots, 58–78
- Matplotlib, 51–53
- plot components, 54–58
-
dataframe.plot.bar()
function, 66–67
- datasets
- datasources
- about, 318
- Amazon SageMaker, 356
- creating, 321–332
- test, 330–332
- training, 324–330
- decision trees, 6, 109–113
-
DecisionTreeClassifier
class, 110, 131
- deep learning, 7
- deleting
- objects, 196
- Python AWS Lambda functions using AWS Management Console, 253–254
- Dense Neural Network (DNN), 390, 408
-
deploy()
function, 377, 417
- deploying DNN classifiers using Google TensorFlow Estimators API and Amazon SageMaker, 408–418
-
DESCR
function, 27, 87
-
describe()
function, 39
- descriptive statistics (Amazon Machine Learning), 320–321
- DetectEntities API, 263–264
-
DetectEntities
function, 270
- detecting
- entities with AWS CLI, 263–264
- key phrases with AWS CLI, 264–265
- DetectKeyPhrases API, 264–265
- DetectSentiment API, 265
- developer SDKs, 155–156
- Diaconis, Persi, 61
- dimensionality reduction, 10
-
display.max_columns
property, 34
- DNN (Dense Neural Network), 390, 408
- DNN classifiers, training/deploying using Google TensorFlow Estimators API and Amazon SageMaker, 408–418
-
DNNClassifier
class, 412
- document types, for attributes, 222
- dominant language (Amazon Comprehend), 259
- E
- EBS (Elastic Block Store), 354
- EC2 (Elastic Compute Block) service, 3, 165, 349, 357–362, 408
- edge locations, 153–154
- Elastic Block Store (EBS) service, 354
- Elastic Compute Block (EC2) service, 165, 349, 357–362, 408
- Elastic Compute Cloud (EC2) service, 3
- elasticity, as a characteristic of cloud computing, 136
-
elicit_slot
function, 296
- enterprise identity federation, 163
- entities
- Amazon Comprehend, 257–258
- detecting with AWS CLI, 263–264
- episodic tasks, reinforcement learning for, 11
- error function, 6
-
Estimator
class, 383
-
estimator_fn
function, 411
-
eval_input_fn
function, 412, 413–414
- evaluating
- classification models, 119–131
- machine learning models, 115–132
- model performance, 24–25
- regression models, 115–119
- events, AWS Lambda and, 240–242
- exceptions, AWS Lambda and, 242–243
- execution environment, AWS Lambda and, 243
- execution role, AWS Lambda and, 239
- execution timeout, AWS Lambda and, 239
- F
- face collection, in Amazon Rekognition, 422
- facial recognition, in Amazon Rekognition, 422
- false negative (FN) count, 121
- false positive (FP) count, 121
- feature engineering, 6
-
feature_names
attribute, 28
- features. See input variables
- Federated Identities (Amazon Cognito), 203
-
figure
class, 57
- figure object, 55
-
fillna()
function, 42–43
-
fit()
method, 89, 382–383
-
fit_intercept
parameter, 90
- FN (false negative) count, 121
- FP (false positive) count, 121
- fraud detection, machine learning for, 7
- Freedman, David, 61
- Freedman-Diaconis rule, 61
- free-tier accounts, signing up for, 142–148
- functions
-
ACMEBankBotAccountOverview
, 286–299
-
ACMEBankBotTransactionList
, 299–304
-
auc()
, 126
- AWS Lambda, 238–239, 435–444
-
boxplot()
, 39–40, 72
-
confusion_matrix()
, 123, 128
-
corr()
, 40
- creating in AWS Lambda, 285–304
-
cut()
, 44–45
-
dataframe.plot.bar()
, 66–67
-
deploy()
, 377, 417
-
DESCR
, 27, 87
-
describe()
, 39
-
DetectEntities
, 270
-
elicit_slot
, 296
- error, 6
-
estimator_fn
, 411
-
eval_input_fn
, 412, 413–414
-
fillna()
, 42–43
-
generate_scatter_plot
, 76
-
head()
, 34, 35, 80, 393
-
hist()
, 37, 59
-
info()
, 32, 33
-
isnull()
, 33
-
keras_model_fn
, 412
-
load_boston()
, 29
-
load_breast_cancer()
, 30
-
load_diabetes()
, 29
-
load_digits()
, 29
-
load_iris()
, 29
-
load_linnerud()
, 29
-
load_wine()
, 30
-
make_blobs
, 77
-
map()
, 413
-
model_fn
, 412
-
NumPy
s, 53
-
pie()
, 69
-
plot.pie()
, 69
-
plot.scatter()
, 74
-
predict()
, 94, 377, 417, 418
-
print()
, 94, 106, 118, 119, 126, 129, 418
-
pyplot hist()
, 59
-
pyplot pie()
, 69
-
qcut()
, 45
-
roc_curve()
, 124–125
-
scatter_matrix()
, 41, 77
-
serving_input_fn
, 412
-
set_index()
, 35
- sigmoid, 23–24, 102
-
sklearn.metrics.mean_r2_score()
, 119
-
sklearn.metrics.mean_squared_error()
, 118
- softmax, 107
-
sum()
, 33
-
suptitle()
method/, 57
-
tf.constant()
, 388
-
tf.feature_column.numeric_column()
, 412
-
tf.get_variable()
, 388, 395
-
tf.global_variables_initializer()
, 400
-
tf.placeholder()
, 388, 394, 395
-
tf.reduce:mean()
, 397
-
title()
, 58
-
train_input_fn
, 412, 413
-
train_test_split()
, 80–82
-
validate_customer_identifier()
, 298–299
-
value_counts()
, 36, 37
-
xlabel()
, 56
-
ylabel()
, 56
- G
- GD (gradient descent) optimization, 397
-
generate_scatter_plot
function, 76
-
get_n_splits()
parameter, 85
- Gini scores, 112
- The Gini Methodology: A Primer on Statistical Methodology (Yitzhaki and Schechtman), 112
- global infrastructure (AWS)
- about, 151
- accessing AWS, 154–160
- edge locations, 153–154
- regions and availability zones, 151–153
- global secondary index, 223
- global tables, in Amazon DynamoDB, 222
- Google Authenticator, 178
- Google TensorFlow
- about, 387–390
- creating linear regression models with, 390–408
- training/deploying DNN classifiers using TensorFlow Estimators API and Amazon SageMaker, 408–418
- using Amazon SageMaker with, 387–418
- gradient descent (GD) optimization, 397
- Graphviz, 111
-
grid()
method, 57
- grids, in plots, 57
-
GridSearchCV
class, 131
- grouped bar charts, 62–65
- groups
- about, 163–164
- modifying permissions associated with existing, 172–173
- H
- handler name, AWS Lambda and, 239
- handlers, AWS Lambda and, 239
- handling missing values, 42–44
- Handwritten digits dataset, 29
- hardware MFA device, 178
- hardware virtualization, 136
- HCA (Hierarchical Cluster Analysis), 9
-
head()
function, 34, 35, 80, 393
- hidden layer, 389
- Hierarchical Cluster Analysis (HCA), 9
-
hist()
function, 37, 59
- histograms, 58–62
- Home menu (AWS management console), 157
- Human Genome Project, 30
- hybrid cloud, 139
- hyperparameter values, selecting, 131–132
- I
- IaaS (infrastructure as a service), 136, 137
- IAM Policy Language, 169
- Identity and Access Management (IAM)
- about, 161
- common tasks, 165–180
- key concepts, 161–165
- identity federation, 162–163
- identity pool (Amazon Cognito)
- about, 203
- compared with user pool, 218–219
- creating, 214–217
- identity provider, 202
- images
- analyzing using Amazon Rekognition Management Console, 423–428
- analyzing with AWS CLI, 428–433
- incremental learning, 12
- index, creating in Amazon DynamoDB, 231–233
-
info()
function, 32, 33
- infrastructure as a service (IaaS), 136, 137
- input data, 318
- input features, selecting, 18–22
- input layer, 389
- input variables, 5
- creating new, 44–46
- one-hot encoding categorical, 47–50
- instance types, 355
- instance-based learning, 12
- insurance premium calculation, machine learning for, 7
- intent (Amazon Lex), 276
- An Introduction to Support Vector Machines and Other Kernel-based Learning Methods (Cristianini and Shawe-Taylor), 96
- IPython Notebook, 4
- Iris plants dataset, 29
-
isnull()
function, 33
- items
- adding to tables in Amazon DynamoDB, 228–231
- in Amazon DynamoDB, 222
- J
- Jason Web Token (JWT)-based identity token, 202
- Jenkins, 238
- Jupyter Notebook, 4
- K
-
Kaggle.com
datasets, 30
-
keras_model_fn
function, 412
- key phrases
- Amazon Comprehend, 258
- detecting with AWS CLI, 264–265
- k-fold cross validation, 84–86
-
KMeans
class, 382–383
- k-means clustering, 9
- L
- L1/L2 regularization, 319
- labeled data, 17
-
lambda_handler()
method, 295
- language support, Amazon Comprehend and, 259
- language-specific SDK, 354
- Latent Dirichlet Allocation (LDA)-based model, 10, 259
- LDA (Linear Discriminant Analysis), 10
- leaf nodes, 388
- Linear Discriminant Analysis (LDA), 10
- linear regression
- about, 23
- creating machine learning models with, 86–92
- creating models with Google TensorFlow, 390–408
- defined, 6
-
LinearRegression
class, 87
- Linnerud dataset, 29
-
load_boston()
function, 29
-
load_breast_cancer()
function, 30
-
load_diabetes()
function, 29
-
load_digits()
function, 29
-
load_iris()
function, 29
-
load_linnerud()
function, 29
-
load_wine()
function, 30
- local secondary index, 223
- location, 318
- logging, AWS Lambda and, 242
- logistic regression
- about, 23
- creating machine learning models with, 101–109
- defined, 6
- M
- machine learning
- about, 3–4
- real-world applications of, 7
- terminology for, 5–7
- tools for, 4–5
- traditional approach versus, 13–25
- types of systems for, 8–12
- machine learning application services, 140–141
- machine learning as a service (MLaaS), 138
- machine learning datasets. See datasets
- machine learning models
- defined, 5
- evaluating, 115–132
- machine learning platform services, 141–142
- maintenance, AWS Lambda and, 238
-
make_blobs
function, 77
- management console (AWS), 156–160
-
map()
function, 413
- MATLAB, 51
- Matplotlib, 5, 51–53
- Maven, 238
- maximum model size parameter, 319–320
- maximum number of training passes parameter, 320
- McCulloch, Warren, 388
- measured service, as a characteristic of cloud computing, 135
- MFA (multifactor authentication), securing root account with, 176–179
-
MinMaxScaler
class, 46
- missing values, handling, 42–44
- ML models
- about, 318–319
- creating, 337–341
- MLaaS (machine learning as a service), 138
- model-based learning, 12
-
model_fn
function, 412
- models
- building with Amazon Machine Learning, 319
- evaluating performance of, 24–25
- versioning in Amazon Rekognition, 423
- modifying
- permissions associated with existing groups, 172–173
- storage classes of objects, 195–196
- multi-class classification models, 126–131, 318
- multifactor authentication (MFA), securing root account with, 176–179
- munge, 18
- music recommendations, machine learning for, 7
- N
- Natural Language Processing (NLP), 140–141, 257–259
- Natural Language Understanding (NLU), 141
- neural networks, 6–7
- neurons, 389
- NLP (Natural Language Processing), 140–141, 257–259
- NLU (Natural Language Understanding), 141
- non-storage operations, in Amazon Rekognition, 423
- notebook instances (Amazon SageMaker)
- creating, 357–362
- training Scikit-learn models on, 364–368
-
n_splits
parameter, 85
- numeric features, transforming, 46–47
- NumPy, 5
-
NumPy
functions, 53
- O
- OAuth 2.0, 202
- object key, 182
- object location, in Amazon Rekognition, 422
- object metadata (Amazon S3), 184–185
- object value, 182
- objects
- accessing, 191–195
- changing storage classes of, 195–196
- deleting, 196
- detecting in Amazon Rekognition, 421
- observation, 318
- obtaining data, 31–41
- On Demand Mode, 224
- on-demand self-service, as a characteristic of cloud computing, 135
- one-hot encoding categorical features, 47–50
- one-versus-one (OVO) approach, 104–105
- one-versus-rest (OVR) approach, 104
-
OneVsOneClassifier
class, 105
-
OneVsRestClassifier
class, 105
- Onset of diabetes dataset, 29
- OpenID Connect, 202
- operation nodes (op nodes), 388
- output layer, 389
- OVO (one-versus-one) approach, 104–105
- OVR (one-versus-rest) approach, 104
- P
- PaaS (platform as a service), 137
- Pandas, 5
- partition key, 222–223
- partition key and sort key, 223
- password rotation policy, setting up for IAM, 179–180
- PCA (Principal Component Analysis), 10
- performing
- queries in Amazon DynamoDB, 235–236
- scans in Amazon DynamoDB, 233–235
- permissions, modifying, 172–173
- pie charts, 69–71
-
pie()
function, 69
- Pillow, 5
- Pinkham, Chris, 139
- Pitts, Walter, 388
- placeholder leaf nodes, 388
- platform as a service (PaaS), 137
-
plot.pie()
function, 69
- plots
- common, 58–78
- components of, 54–58
-
plot.scatter()
function, 74
-
plt.bar()
statement, 67
- policy, 164
-
predict()
function, 94, 377, 417, 418
- prediction endpoints (Amazon SageMaker), 355
- prediction instances (Amazon SageMaker), 355
- predictions, making using AWS CLI, 347–349
-
predict_proba()
method, 103, 106, 107
- preparing
- test data in Amazon SageMaker, 362–364
- test sets, 22
- training data in Amazon SageMaker, 362–364
- training sets, 22
- pricing
- Amazon Comprehend and, 259–260
- Amazon Lex, 278
- Amazon Machine Learning, 321
- of Amazon Rekognition, 423
- Amazon SageMaker, 357
- AWS Lambda and, 244
- primary keys (Amazon DynamoDB), 222–223
- Principal Component Analysis (PCA), 10
-
print()
function, 94, 106, 118, 119, 126, 129, 418
-
print()
statement, 59–60
- private cloud, 138
- product recommendations, machine learning for, 7
- programmatic access, 168
- programming mode (Amazon SageMaker), 354
- programming model
- Amazon Lex, 277–278
- AWS Lambda and, 239–243
- Provisioned Mode, 225
- PubChem, 30
- public cloud, 138
-
pyplot hist()
function, 59
-
pyplot
module, 51–53
-
pyplot pie()
function, 69
- Python
- about, 4
- creating Lambda functions using AWS Management Console, 244–250
- deleting Lambda functions using AWS Management Console, 253–254
- testing Lambda functions using AWS Management Console, 250–253
- Python, data visualization in
- about, 51
- common plots, 58–78
- Matplotlib, 51–53
- plot components, 54–58
- Q
-
qcut()
function, 45
- queries (Amazon DynamoDB), 223, 235–236
- R
- R language, 4
- R2 metric, 119
-
raise
statement, 242–243
-
random_state
parameter, 80, 85
- read consistency (Amazon DynamoDB), 224
- read/write capacity modes (Amazon DynamoDB), 224–225
- real-time prediction endpoints
- creating, 346–347
- using, 349–350
- receiver operating characteristics (ROC), 124
- Reduced Redundancy Storage (RRS), 182
- regions (AWS), 151–153
- Regions menu (AWS management console), 160
- regression
- defined, 6
- machine learning models used for, 318
- regression models, evaluating, 115–119
- regularization, 319
- regularization amount parameter, 320
- regularization type parameter, 320
- reinforcement learning, 11
- requests, Amazon S3 costs for, 183
- Resource Groups menu (AWS management console), 157–158
- resource pooling, as a characteristic of cloud computing, 135
- resource-based policy, 164
- REST APIs, 218–219
- RESTful web services, 155
- retrieving Amazon Cognito client secret, 213–214
- RMSE (root mean squared error), 87, 117–119
- ROC (receiver operating characteristics), 124
-
roc_curve()
function, 124–125
- roles
- about, 164–165
- creating, 173–176
- root account
- about, 161
- securing with MFA, 176–179
- root mean squared error (RMSE), 87, 117–119
- RRS (Reduced Redundancy Storage), 182
-
run()
method, 400
- S
- SaaS (software as a service), 137
- SAML 2.0 Federation, 174
- Samuel, Arthur (scientist), 4
- scalar types, for attributes, 222
- scans (Amazon DynamoDB), 223–224, 233–235
- scatter plots, 73–78
-
scatter_matrix()
function, 41, 77
- scene detection, in Amazon Rekognition, 422
- Schechtman, Edna (author)
- The Gini Methodology: A Primer on Statistical Methodology, 112
- schema, 318
- Scikit-learn
- about, 4, 79–80
- creating machine learning models with, 79–113
- creating training and test datasets, 80–86
- datasets for, 27–30
- decision trees, 109–113
- k-fold cross validation, 84–86
- linear regression, 86–92
- logistic regression, 101–109
- support vector machines, 92–101
- training models on Amazon SageMaker notebook instances, 364–368
- training models on dedicated training instances, 368–378
- Seaborn, 52
- secondary indexes (Amazon DynamoDB), 223
- securing root account with MFA, 176–179
- selecting
- algorithms, 23–24
- hyperparameter values, 131–132
- input features, 18–22
- semi-supervised learning, 10–11
- sentiment (Amazon Comprehend), 258
- sentiment analysis, with AWS CLI, 265
- serverless back end, AWS Lambda and, 237
- service limitations, AWS Lambda and, 244
- Services menu (AWS management console), 157
-
serving_input_fn
function, 412
- set types, for attributes, 222
-
set_index()
function, 35
-
set_title()
method, 58
-
set_xlabel()
method, 56
-
set_ylabel()
method, 56
- Shawe-Taylor, John (author)
- An Introduction to Support Vector Machines and Other Kernel-based Learning Methods, 96
-
shuffle
parameter, 85
- shuffle type parameter, 320
- sigmoid function, 23–24, 102
-
sklearn.metrics.mean_r2_score()
function, 119
-
sklearn.metrics.mean_squared_error()
function, 118
- slot (Amazon Lex), 276–277
-
SM_CHANNEL_TRAINING
variable, 371
-
SM_MODEL_DIR
variable, 371
-
SM_OUTPUT_DATADIR
variable, 371
- softmax function, 107
- software as a service (SaaS), 137
-
split()
parameter, 85
- stacked bar charts, 65–67
- stacked percentage bar charts, 67–68
- Standard - IA storage class, 182
- Standard storage class, 182
-
StandardScaler
class, 46
- stochastic gradient descent, 402
- storage
- Amazon S3 costs for, 183
- changing for objects, 195–196
- storage class, 182–183
- Storage Management Pricing (Amazon S3), 183
- storage-based operations, in Amazon Rekognition, 423
-
stratify
parameter, 82
- streams, AWS Lambda and, 238
- subresources (Amazon S3), 183
-
sum()
function, 33
- supervised learning, 8–9
- Support menu (AWS management console), 160
- support services, 142
- support vector machine (SVM), 92–101
- supported languages, for AWS Lambda, 238
-
suptitle()
method/function, 57
- SVM (support vector machine), 92–101
- syntax (Amazon Comprehend), 258
- system-defined metadata (Amazon S3), 184
- T
- tables
- adding items to in Amazon DynamoDB, 228–231
- Amazon DynamoDB, 433–435
- in Amazon DynamoDB, 222
- creating in Amazon DynamoDB, 225–228, 278–285, 433–435
-
target
attribute, 28
- target attributes, 318
- target variable, 6
-
target_names
attribute, 28
- tasks (IAM), 165–180
- TensorFlow, 5
-
TensorFlow
class, 415
- tensors, 387
- terminology, 5–7
- test data
- defined, 6
- preparing in Amazon SageMaker, 362–364
- test datasets, creating, 80–86
- test datasources, creating, 330–332
- test sets, preparing, 22
- testing
- bots, 314–315
- Python AWS Lambda functions using AWS Management Console, 250–253
-
test_size
parameter, 80
- text analysis
- with AWS CLI, 262–265
- using Amazon Comprehend Management Console, 260–262
-
tf.constant()
function, 388
-
tf.feature_column.numeric_column()
function, 412
-
tf.get_variable()
function, 388, 395
-
tf.global_variables_initializer()
function, 400
-
tf.GradientDescentOptimizer
class, 398
-
tf.matmul()
operator, 396
-
tf.placeholder()
function, 388, 394, 395
-
tf.reduce:mean()
function, 397
-
tf.reset_default_graph()
statement, 395
-
tf.square()
operator, 397
-
tf.zeros()
operator, 395
- Titanic dataset, 31–50, 321–332
- title, of plots, 57–58
-
title()
function, 58
- TN (true negative) count, 121
- tools, 4–5
- topic modeling, Amazon Comprehend and, 259
-
torrent
object, 183
- TP (true positive) count, 121
- traditional approach, machine learning versus, 13–25
- training
- datasources, 324–330
- defined, 5
- DNN classifiers using Google TensorFlow Estimators API and Amazon SageMaker, 408–418
- models using built-in algorithms on dedicated training instances in Amazon SageMaker, 379–384
- Scikit-learn models on dedicated training instances in Amazon SageMaker, 368–378
- Scikit-learn models on notebook instances in Amazon SageMaker, 364–368
- training data
- defined, 6
- preparing in Amazon SageMaker, 362–364
- training datasets, creating, 80–86
- training datasources, creating, 324–330
- training instances
- training models using built-in algorithms on dedicated, 379–384
- training Scikit-learn models on dedicated, 368–378
- training jobs (Amazon SageMaker), 354–355
- training parameters (Amazon Machine Learning), 319–320
- training sets, preparing, 22
-
train_input_fn
function, 412, 413
-
train_test_split()
function, 80–82
- Transfer Acceleration, Amazon S3 costs for, 183
- transforming numeric features, 46–47
- triggers
- Amazon Cognito, 212–213
- AWS Lambda, 237
- true negative (TN) count, 121
- true positive (TP) count, 121
- Tukey, John, 71
- U
- U2F Key, 177
- UCI machine learning repository, 30–31
- unsupervised learning, 9–10
- user pool (Amazon Cognito)
- about, 202–203
- compared with identity pool, 218–219
- creating, 204–213
- user-based policy, 164
- user-defined metadata, 185
- users
- utterance (Amazon Lex), 277
- V
-
validate_customer_identifier()
function, 298–299
-
value_counts()
function, 36, 37
- variable leaf nodes, 388
- version ID, 182
- video recommendations, machine learning for, 7
- viewing data insights, 332–337
-
ViewTransactionList
intent, customizing, 312–314
- virtual MFA device, 177
- W
- Web Identity, 174
- web identity federation, 163
- websites
- Amazon Comprehend, 259, 260, 270
- Amazon DynamoDB, 221, 225
- Amazon Lex, 278
- Amazon Machine Learning, 321, 336
- Amazon Machine Learning and AWS CLI, 347
- Amazon Polly, 141
- Amazon Rekognition, 423, 439
- Amazon SageMaker pricing/availability, 357
- Amazon Transcribe, 141
- Amazon Translate, 141
- Anaconda Navigator, 4
- attribute data types, 222
- availability zones (AZ), 152–153
- AWS APIs, 262
- AWS DeepLens, 142
- AWS Lambda pricing, 244
- AWS services, 240, 317
- AWS Signature V4, 417–418
- batch predictions with Amazon SageMaker, 355
- batch transforms, 376, 416
- building custom Estimators, 411
- built-in algorithms, 356
- chatbots and third-party messaging platforms, 314
- creating feature columns, 412
-
DNNClassifier
class, 412
- Docker registries for Scikit-learn Docker images, 368
- Docker registries for TensorFlow Docker images, 408
- environment variables, 371
-
Estimator
class, 383
- gradient descent technique, 397
- IAM Policy Language, 169
- IDEs and tools for creating Lambda code, 238
- Jupyter Notebook, 4
- Kaggle.com, 30–31
-
KMeans
class, 382
- language detection APIs, 259
- Matplotlib, 5
- monitoring Amazon Rekognition, 427
- NumPy, 5
- OpenID Connect specifications, 207
- Pandas, 5
- Pillow, 5
- prediction instance types, 355
- PyTorch, 5
- real-time prediction API, 350
- regions (AWS), 152–153
- sample event data, 242
- Scikit-learn, 4
- serving models with TensorFlow Serving, 414
-
SKLearn
class parameters, 374
- standard libraries with containers, 243
- TensorFlow, 5
- TensorFlow APIs, 390
-
TensorFlow
class, 415
-
tf.GradientDescentOptimizer
class, 398
-
tf.matmul()
operator, 396
-
tf.placeholder()
function, 395
-
tf.zeros()
operator, 395
- Titanic dataset, 321
- Titanic Machine Learning From Disaster, 321
- training instance types, 355
- Wikipedia, 30
- Wine recognition dataset, 30
-
with-as
statement, 400
- Y
- Yitzhaki, Shlomo (author)
- The Gini Methodology: A Primer on Statistical Methodology, 112
..................Content has been hidden....................
You can't read the all page of ebook, please click
here login for view all page.