Date post: | 15-Jan-2017 |
Category: |
Technology |
Upload: | sunview-software-inc |
View: | 217 times |
Download: | 0 times |
+
Hosted By:John PrestridgeVP of Marketing & Product StrategySunView Software
Guest Speaker:Lawrence O. HallDistinguished University ProfessorDept. of Computer Science& EngineeringUniversity of South Florida
How Big Data &Machine Learning are Transforming ITSM
Today’s Presenters
2
John Prestridge - HostVP of Marketing and Product Strategy – SunView Software
Lawrence O. Hall – Guest SpeakerDistinguished ProfessorDept. of Computer Science & EngineeringUniversity of South Florida
Housekeeping
3
• This webinar will be available shortly after its conclusion• Share this webinar and check out the supplemental resource kit for
machine learning and ITSM• Have a question regarding anything that is covered during this
webinar? Use the BrightTalk ‘Ask A Question’ window to submit your question to the webinar panel!
Agenda
4
An Overview of Big Data & Machine Learning
Big Data & Machine Learning for ITSM
Q&A
5
Key Drivers of Machine Learning
6
Lawrence O. Hall – Guest SpeakerDistinguished ProfessorDept. of Computer Science & EngineeringUniversity of South Florida
7
What is Big Data?
Data that has a large:• Volume: lots of records
• Variety: lots of different kinds of data
• Velocity: changing fast
• Or some combination of the three V’s
8
Big Data
We Need New Ways to Analyze Data
Curating and storing lots of data can be a challenge when using machine learning for predictive analytics.
9
Big Data Examples
• The friends network in Facebook
• Amazon history of purchases
• Records of cell phone calls, texts, or tweets
• History of all service requests in your company
10
Big Data Examples
• There is velocity as posting is constant
• Evenings lead to more posts
• That cute cat picture gets posted…. A lot!
Consider image posts and shares on Facebook
Netflix records ratings of movies and shows by user
11
What Do We Want to Know from Big Data?
• Amazon wants to suggest books you might buy• Or, perhaps Amazon suggests related material based
of a past purchase of biking gloves…• How do they do this?
12
Machine Learning
• What is machine learning, data mining, and predictive analytics?
• From data, preferably with some ‘class’ labels, a machine learning algorithm can build a predictive model
• Amazon, has lots of users and products. If it can aggregate what users have bought together (or over time), it can suggest what you might like to buy
13
Are All Those Cats The Same?
• If a learned model can recognize the same image, Facebook and others who store images can have just one linked copy
• There are now models that are nearly perfect at matching the same thing
14
Machine Learning/Data Mining AlgorithmsThere are many algorithms and we only touch on a few
• Decision tree algorithms are fast to build and reasonably accurate. Use a random forests ensemble for better accuracy with a wisdom of crowds approach
• If you have big, labeled image data – Convolutional Neural Networks, using deep learning, are really good
• Support Vector Machines let you project data into a higher dimension (“kernel trick”) and then linearly separate them
• No labels but you want to group data? Try K-means or fuzzy K-means clustering
15
Decision Tree Example
For our data, we need features.
Assume we want to decide whether to play tennis and have historical data…
16
Decision Tree Example -Evaluating Weather Attributes
Outlook Temp Humidity Windy Play
Sunny Hot High False No
Sunny Hot High True No
Overcast Hot High False Yes
Rainy Mild High False Yes
Rainy Cool Normal False Yes
Rainy Cool Normal True No
Overcast Cool Normal True Yes
Sunny Mild High False No
Sunny Cool Normal False Yes
Rainy Mild Normal False Yes
Sunny Mild High True No
Overcast Mild High True Yes
Overcast Hot Normal False Yes
Rainy Mild High True No
17
Decision Tree Example -Evaluating Weather Attributes
Attribute Rules Errors Total errors
Outlook Sunny No 1/5 3/14
Overcast Yes 0/4
Rainy Yes 2/5
Outlook Temp Humidity Windy Play
Sunny Hot High False No
Sunny Hot High True No
Overcast Hot High False Yes
Rainy Mild High False Yes
Rainy Cool Normal False Yes
Rainy Cool Normal True No
Overcast Cool Normal True Yes
Sunny Mild High False No
Sunny Cool Normal False Yes
Rainy Mild Normal False Yes
Sunny Mild High True No
Overcast Mild High True Yes
Overcast Hot Normal False Yes
Rainy Mild High True No
18
Decision Tree Example -Evaluating Weather Attributes
Attribute Rules Errors Total errors
Outlook Sunny No 1/5 3/14
Overcast Yes 0/4
Rainy Yes 2/5
Temp Hot No* 2/4 6/14
Mild Yes 3/6
Cool Yes 1/4
Humidity High No 3/8 4/14
Normal Yes 1/6
Windy False Yes 2/8 4/14
True No* 2/6
* indicates a tie
Outlook Temp Humidity Windy Play
Sunny Hot High False No
Sunny Hot High True No
Overcast Hot High False Yes
Rainy Mild High False Yes
Rainy Cool Normal False Yes
Rainy Cool Normal True No
Overcast Cool Normal True Yes
Sunny Mild High False No
Sunny Cool Normal False Yes
Rainy Mild Normal False Yes
Sunny Mild High True No
Overcast Mild High True Yes
Overcast Hot Normal False Yes
Rainy Mild High True No
19
Decision Tree Example -Best First Test
Temp Humidity Windy Play
Hot High False NoHot High True NoMild High False NoCool Normal False YesMild High True No
Sunny
Overcast
Rainy
3 - Yes
3-Yes2 - No
Outlook
20
Decision Tree Example -Best First Two Tests
Sunny
Overcast
Rainy
3 - Yes
HighNormal
4 - No1 - Yes
3-Yes2 - No
Outlook
Humidity
21
Decision Tree Example -Final Tree
Sunny
Overcast
Rainy
3 - YesHumidity
High Normal
4 - No1 - Yes
TrueFalse
2 - No 3 - Yes
Outlook
Windy
22
• Now you know something of big data
• You have heard of some machine learning success
• You can build a simple decision tree!
23
John PrestridgeVP of Marketing and Product Strategy
24
25
Supporting the Digital Workplace
BYOD
CLOUD
MOBILEWORKFORCE
SHADOW IT
IOT
Volume , Velocity and Variety of Requests Business will expect more apps, delivered
more quickly, with consumer-like support Do more with less
SELF-SERVICE
26
Supporting the Digital Workplace
Transition from being reactive to a proactive delivery of services that
leverages a people-centric approach to empower employee effectiveness.
27
Key Opportunity
-
By 2019, IT service desks utilizing machine-learning enhanced technologies will free up to 30% of support capacity.*
*Apply Machine Learning and Big Data at the IT Service Desk to Support the Digital Workplace February 2016 Analyst(s): Colin Fletcher | Katherine Lord
28
Big Data + Machine Learning
DATA
Ticket HistoryKnowledgeAssetsInteractionsUsage Patterns…..
Large ScaleData ProcessingEnvironment
90% of data today is machine generated
or people interactions
DOMAINMODEL
MACHINELEARNING
Algorithms Regression Anomaly Detection Clustering Classification ....
29
Big Data + Machine Learning
DATA
Ticket HistoryKnowledgeAssetsInteractionsUsage Patterns…..
DOMAINMODEL
MACHINELEARNING
30
Big Data + Machine Learning
DATA
Ticket HistoryKnowledgeAssetsInteractionsUsage Patterns…..
IncidentService RequestProblemChange…..Algorithms
Regression Anomaly Detection Clustering Classification ....
DOMAINMODEL
MACHINELEARNING
31
Big Data + Machine Learning
DATA
Ticket HistoryKnowledgeAssetsInteractionsUsage Patterns…..
IncidentService RequestProblemChange…..Algorithms
Regression Anomaly Detection Clustering Classification ....
NEEDS: ITSM Expert Data Scientist Big Data Infrastructure Machine Learning Tools
DOMAINMODEL
MACHINELEARNING
32
Big Data + Machine Learning
DATA
Ticket HistoryKnowledgeAssetsInteractionsUsage Patterns…..
INTELLIGENTFEATURES
Recommendation Engines Intelligent Search Predictive Analytics BOTS
API
Algorithms Regression Anomaly Detection Clustering Classification ....
33
ITSM + Machine Learning
IntelligentFeatures
Predictive Analytics User Sentiment Score Change Risk Predict Problems
Intelligent Search Knowledge Curation Smart Notifications Contextual Search
BigData
MachineLearning
Recommendation Engines Resolution Suggestions Level 1 Ticket Completion Intelligent Routing
“Better Decisions” “Faster Resolutions”
“Improved Self-Service”“Engaged Users”
BOTS Intelligent Autoresponder Self-Service Virtual Assistant
34
Summary
Big Data is here and Machine Learning is aproven technology
Need proactive delivery of services to support the digital workplace
Invest in big data, machine learning, and other AI technologies to transform ITSM
35
ITSM + MACHINE LEARNING
www.sunviewsoftware.com/learn/machine_learningLearn More:
36
Get Connected
Do you have any personal experience or additional questions regarding the topics we covered today?
Get into the discussion via email:• Lawrence Hall: [email protected]• John Prestridge: [email protected]
37
Q&A
Thank You!If you would like to find out more visitwww.SunViewSoftware.com
LinkedIn.com/companies/sunview-software-inc-Twitter.com/SunViewSoftwareFacebook.com/SunViewSoftware