You are on page 1of 24

Oracle Machine Learning on Cloud

Objectives
Illustrate Machine Learning

Features of Oracle Machine Learning

Describe the Oracle Machine Learning process

Application areas of Oracle Machine Learning

Know the different types of Machine Learning Algorithms

Describe the Notebooks in Oracle Machine Learning

Use Case: Predicting the Insurance Buyers


Machine Learning

• Machine learning is a subset of artificial intelligence.


• Machine Learning algorithms are trained with the help of training data
set to create a model. Machine learning
is truly the future
• The algorithms use the patterns available in the training data to of analytics in
every industry.
perform classification and predictions.
• Oracle Machine Learning:
– Is a collaborative web-based interface
– Provides a development environment to create data mining notebooks
– Enables data analytics, data discovery, and data visualizations
– Is compatible and is deployed along with Oracle Autonomous Database
services
3
Oracle Machine Learning with Autonomous Database/DW

Oracle Machine
Learning

4
Features of Oracle Machine Learning

• Built-in Web-based SQL Machine Learning Tool


• Packaged with Oracle Autonomous Data Warehouse, Oracle Autonomous Transactional
Processing, and Autonomous Oracle Analytics Cloud
• Easy access to shared notebooks, templates, permissions, and schedulers
• Access to 30+ parallel, scalable, in-database implementations of machine learning algorithms
• SQL and PL/SQL scripting language support
• Support for deployments of Enterprise Machine Learning methodologies in ADW

5
Oracle Machine Learning Process

Step 1 Step 2 Step 3 Step 4 Step 5

Data Collection
Data Visualization-
from
Heterogenous Build &Train the Transforming the
Data Preparation
Sources Model Make Predictions results into visual
graphs

6
Application Areas of Oracle Machine Learning

• Static Image Recognition


• Video Surveillance
• Financial services
• Healthcare and life sciences
• Oil and gas
• Transportation
• Retail
• Sentiment Analysis
• Speech Recognition
• Online Fraud Detection
7
Types of Machine Learning Algorithms
Machine Learning algorithms are divided into four types:

Machine Learning

Supervised Learning Unsupervised Learning Semi-supervised Learning Reinforcement Learning

Incoming data is labeled based on Discovers through trial


It maps an input to an output
past data samples & manually trains and error and decides
based on example input-output Combination of Supervised
the algorithm to recognize certain which action results in
pair. and Unsupervised Learning
types of objects and categorize them. greater rewards
8
Oracle Machine Learning Algorithms
X2

CLASSIFICATION ANOMALY TIME


• Naïve Bayes DETECTION X1
SERIES
• Logistic Regression (GLM) • One-Class SVM • State-of-the-art forecasting
• Decision Tree using Exponential
• Random Forest Smoothing
• Neural Network
ASSOCIATION • Includes all popular models
• Support Vector Machine
RULES For example, Holt-Winters
• A Priori/Market Basket with trends, seasons,
• Explicit Semantic Analysis
SQL ANALYTICS irregularity, missing data
• SQL Windows, SQL Patterns,
CLUSTERING SQL Aggregates
• Hierarchical K-Means
• Hierarchical O-Cluster PREDICTIVE QUERIES
• Expectation Maximization (EM) • Predict, Cluster, Detect, Features

9
Oracle Machine Learning Algorithms
REGRESSION STATISTICAL FEATURE
• Linear Model FUNCTIONS EXTRACTION
• Generalized Linear Model • Basic Statistics: min, max, • Principal Comp Analysis (PCA)
• Support Vector Machine (SVM) median, stdev, t-test, F- • Non-Negative Matrix Factorization
test, Pearson’s, Chi-Sq,
• Stepwise Linear Regression • Singular Value Decomposition
ANOVA, and so on
• Neural Network (SVD)
• LASSO * R PACKAGES • Explicit Semantic Analysis (ESA)
• CRAN R Algorithm
Packages Through
ATTRIBUTE Embedded R Execution TEXT MINING
IMPORTANCE A1 A2 A3 A4 A5 A6 A7 • Spark MLlib Algorithm SUPPORT
Integration • Algorithms Support Text
• Minimum Description Length
• Tokenization and Theme Extraction
• Principal Comp Analysis (PCA) EXPORTABLE • Explicit Semantic Analysis (ESA) for
• Unsupervised Pair-Wise KL Div ML MODELS Document Similarity
• CUR Decomposition for Row & AI • REST APIs for Deployment

10
Notebooks in Oracle Machine Learning

• A notebook is a web-based interface for data analysis, data discovery, data visualization, and
collaboration.
• Oracle Machine Learning SQL notebooks provide easy access to Oracle’s parallelized, scalable,
in-database implementations of a library of Oracle Advanced Analytics’ machine learning
algorithms (classification, regression, anomaly detection, clustering, associations, attribute
importance, feature extraction, times series, and so on.)
• Oracle Machine Learning uses Zeppelin Notebook, a collaborative interface to write the code,
equations, and text, create visualizations, and perform data analytics.

12
OML Use Case: Predicting the Insurance Buyers

• The Notebook shows predicting the


customers most likely to buy insurance based
on the customer data using the supervised
learning algorithm.
• The Notebook first builds a (classification)
decision tree model, evaluates and applies the
model to find the target customers.

13
Oracle Machine Learning Use Case: Example
OMLUSER1
Workspace Name

OML Notebook
Name

Use Case
Description

14
Oracle Machine Learning Use Case: Example

Fetch data from the Customer Information


INSUR_CUST_LTV_SAMPLE
table.

15
Oracle Machine Learning Use Case: Example

Running the Cell in


Notebook
Drop the Previous Classification
Model (Decision tree)

Drop the previously applied


result

16
Oracle Machine Learning Use Case: Example
Run the Notebook

Building the Model using


Classification Decision Tree
Algorithm

Train and Test the Model

Script result
17
Oracle Machine Learning Use Case: Example
Display the Cumulative Gain Result
(QUANTILE_NUMBER,GAIN_CUMUL
ATIVE)

18
Oracle Machine Learning Use Case: Example

Displaying the Apply result in Model

19
Oracle Machine Learning Use Case :Example

Displaying the Customers above


Stated Threshold Probability of
Buying Insurance

20
Oracle Machine Learning Use Case: Example

Display the customers who are


likely to buy insurance

21
Oracle Machine Learning Use Case: Example

Display the region-wise top nine


customers who are likely to buy
insurance

22
Oracle Machine Learning Use Case: Example

Display the customers above a


Stated Threshold Probability of
buying insurance

23
Oracle Machine Learning Use Case: Example

"Score on the Fly" prediction for a


particular customer's data record

• Successfully predicted the customers most likely to buy insurance using the classification
(decision tree) model based on the customer “INSUR_CUST_INSUR_LTV”data

24
Summary
Illustrate Machine Learning

Features of Oracle Machine Learning

Describe the Oracle Machine Learning process

Application areas of Oracle Machine Learning

Know the different types of Machine Learning Algorithms

Describe the Notebooks in Oracle Machine Learning

OML Use Case: Predicting the Insurance Buyers

You might also like