Predictive analytics-nirmal.potx

Predictive Analytics: From Data to Foresight

Summer School 2015

Nirmal Fernando

Associate Technical Lead, WSO2 Inc.

Predictive Analytics

Extract information from existing data sets to determine patterns and predict future outcomes and trends.

It does not tell you what will happen in the future.

It forecasts what might happen in the future with an acceptable level of reliability.

2

source: http://insidebigdata.com/2014/08/25/salespredict-marketo-partner-using-predictive-analytics/

Where is it used?

3

Ticket to success

o Guiding front-line decisions and

actions via transmuting data into

predictive visions and intuitions.

o Customer requirements and steps to

increase profitability and retention.

o Boosting productivity of people, assets and processes

o Eliminating threats and frauds before they can hamper the

image and reputation of the company.

o Assessing the social media impact of your products in the

market.

4

o Matured

o Less expensive

o More approachable

o Easy to make use of

Unlocking the potential of “Big Data”

5

Machine Learning

Machine learning is the science of getting computers to act without being explicitly

programmed.

6

Categorization

o Supervised Learning

o Unsupervised Learning

o Reinforcement Learning

7

Categorization

8

o Supervised Learning

o Unsupervised Learning

o Reinforcement Learning

Supervised Learning

Machine learning task of inferring a function from labeled training data.

9

source: http://www.astroml.org/sklearn_tutorial/general_concepts.html

Supervised Learning - Classification

10

Supervised Learning - Regression

11

Supervised Learning - Algorithms

12

Regression

o Linear Regression

o Lasso Regression

o Ridge Regression

Classification

o Logistic Regression

o Support Vector Machine

o Decision Tree

o Random Forest

o Naive Bayes

Unsupervised Learning

Machine learning task of inferring a function from unlabeled training data. The algorithm tries to find similarities among the objects in question.

13source: http://www.astroml.org/sklearn_tutorial/general_concepts.html

Unsupervised Learning - Clustering

14

o Manage and explore your data o Analyze the data using machine learning algorithmso Build machine learning modelso Compare and manage generated machine learning modelso predict using the built models

Powered by Apache Spark and Apache Spark MLlib.

Key words,

o ML Project: a logical grouping of set of machine learning analyses you would perform on a selected dataset.

o ML Analysis: holds a pre-processed feature set, a selected machine learning algorithm and its calibrated set of hyper-parameters.

WSO2 Machine Learner

15

Manage your datasets...

16

Explore your data..

17

Manage ML projects...

18

Generate and manage ML models

19

Finding the best model

20

Other features...

21

o Fast and scalable machine learning

o Every operation exposed using a REST API

o Easy to use User Interface

o Use generated models in WSO2 ESB and WSO2 CEP for

prediction

Future,

o Deep learning algorithms

o NLP techniques

o Data pre-processing techniques

Let’s try to solve a real world problem

22

Problem description

23

source: https://www.kaggle.com/c/otto-group-product-classification-challenge

Dataset

24

o 93 features

o for 200,000+ products

o id - an anonymous id unique to a product

o feat_1, feat_2, ..., feat_93 - the various features of a product

o target - the class of a product. There are 9 most important

product categories (like fashion, electronics, etc.)

Objective

25

Build a predictive model

which is capable of

categorizing a given

product into

one of the categories.

Towards the solution..

26

27

Summary

28

o Discussed predictive analytics

o Learnt what machine learning is

o Got to know widely-used machine learning techniques

o Glanced at WSO2 Machine Learner product features

o Solved a real-world machine-learning problem using WSO2

Machine Learner

29

Contact us !