![© 2018 Burning Glass Technologies—Proprietary and Confidential
Job Posting Salary Prediction API
20
• Predicting Salary associated with a job posting
• Understanding the impact of different variables in salary prediction
• Geolocation (is NYC more expensive than Boston for the same job
posting)
• Title – Can we understand impact of title (level I vs Level II) in
salary
• Impact of higher value skills in job posting
• Data Scientist1 - [Python, Java, data analysis]
• Data Scientist 2- [Python, Java, ,machine learning, deep
learning]
• Data Analyst 1– [excel, machine learning, python]
• Data Analyst2 – [word, excel, data analysis]
• Return of education and experience for an occupation – Does
salary changes uniformly for all occupation or differently?](https://image.slidesharecdn.com/uassdatasciencepresntation-180515143821/85/workforce-analytics-using-Data-Science-20-320.jpg)
workforce analytics using Data Science
- 1. © 2018 Burning Glass Technologies—Proprietary and Confidential Burning Glass Technologies Data Driven Workforce Analytics Data Science Research Symposium University of Massachusetts - Amherst 12 April 2018 Manish Gaurav Lead Data Scientist mgaurav@burning-glass.com
- 2. © 2018 Burning Glass Technologies—Proprietary and Confidential Agenda • Overview of Burning Glass • About the Data • How we use Data Science • Labor Market Entity Similarity • Labor Market Content Tagging • Salary Compensation Insights 2
- 3. © 2018 Burning Glass Technologies—Proprietary and Confidential Overview of Burning Glass 3
- 4. © 2018 Burning Glass Technologies—Proprietary and Confidential DISRUPTIVE FORCES 4 Automation Skill Hybridization Signal failure Education poorly aligned with demand Workers lack info for making smart career choices Weak talent supply chains Job Market Disruption ARE DRIVING MARKET INEFFICIENCIES ACROSS STAKEHOLDERS
- 5. © 2018 Burning Glass Technologies—Proprietary and Confidential JOBS HAVE A GENOME • Jobs are defined by skills • Increasingly, the unit of currency of the job market is skills – not jobs • Skills express the job market’s dynamism • Skills are the key to unlocking mobility within and between roles • Mapping between skills and jobs provides a powerful bridge between education & work AND THE WORLD ECONOMY NEEDS A MAP
- 6. © 2018 Burning Glass Technologies—Proprietary and Confidential Burning Glass at a Glance • 320 employees across four continents • 40+ analysts, data scientists and researchers • 500+ customers • Leadership with deep expertise in text analytics and in HCM, education & public policy • >60% of revenue reinvested in R&D Burning Glass is an analytics software company that leverages the world’s largest and most sophisticated database of jobs and talent to crack the genetic code of the job market, delivering real-time data and breakthrough planning tools that inform careers, define learning & development, and shape workforces. WE TURN REAL-TIME DATA INTO ACTIONABLE INSIGHTS
- 7. © 2018 Burning Glass Technologies—Proprietary and Confidential 80%3.4 million 50,000 300 million >1 billion Dynamic Labor Market Taxonomy >1 million Active unique jobs collected daily Sources across the web - job boards and corporate sites Historical job market records Firms represented, from large corporations to SME’s 23 Career Areas 1700 Occupations 18,000 Skills 60,000 Skill Variants Deduplication ensuring integrity and consistency Resumes processed per annum Burning Glass data have been built over 15 years with robust taxonomies and the industry’s largest in-house data science team. BURNING GLASS BRIDGES DATA SCIENCE AND PRACTICAL APPLICATION
- 8. © 2018 Burning Glass Technologies—Proprietary and Confidential About Our Data 8
- 9. © 2018 Burning Glass Technologies—Proprietary and Confidential •Volume •Emergence •Event based Identify Sources •Job boards •Career pages •Direct feed Aggregate •De-duplicate •Structure •Normalize Pre-process •Reporting •Metrics •Indices Analyze •Actions •Options •Forecasts Insights 9 BURNING GLASS AUTOMATES THE CURATION OF BIG DATA IN HCM Human Supervision Each process informed and supervised by data scientists, economists, policy experts, etc. Machine Supervision Neural network enabled similarity and content recommendation engine for labor market entities Industry Supervision Burning Glass sits at the nexus of policy, academia, and industry; providing key insights on labor trends and guiding quality control Burning Glass combines scale and quality through the marriage of human, machine, and industry supervision.
- 10. © 2018 Burning Glass Technologies—Proprietary and Confidential OCCUPATIONAL ONTOLOGY Occupation Hierarchy Sample: Logistics Metadata Elements • Demand • Projected Growth • Average Salary • Industries Hiring • Employers Hiring • Skills in Demand • Salary Booster Skills • Degree Level Distribution • Similar Jobs • Posting Duration • Description
- 11. © 2018 Burning Glass Technologies—Proprietary and Confidential SKILL ONTOLOGY Skill Hierarchy Sample: Web and Mobile Metadata Elements • Skill Type • Description • Demand • Projected Growth • Occupations Hiring • Average Salary • Industries Hiring • Employers Hiring • Similar Skills
- 12. © 2018 Burning Glass Technologies—Proprietary and Confidential How we use Data Science 12 Similarity Classification Prediction
- 13. © 2018 Burning Glass Technologies—Proprietary and Confidential Labor Market Entity Similarity API 13 • Discover similarities between underlying concepts in the labor market • Extracting relationship between skills and occupations. • Is ‘data science’ more similar to ‘machine learning’ or ‘Microsoft excel’ • Is ‘data scientist’ more similar to ‘data engineer’ or ‘data analyst • What is it • Neural Network powered similarity Model • Understands patterns based on relationship in the labor market
- 14. © 2018 Burning Glass Technologies—Proprietary and Confidential Labor Market Entity Similarity API 14 • Uses a neural network powered BGT Similarity API • Neural network converts word to vectors which can then be used to compute the notion of semantic similarity • vector(“king”) – vector(“man”) + vector(“woman”) ≈ vector(“queen”) • Can compute similarity across skill x occupation • Occupation + skill1,skill2…. => list(occupations) • Data analyst + python,machine learning ~ data scientist (recommendation engine) • Different entities can be combined to understand similarity as well • Can we understand similarity of companies as well ??? • What companies are similar to American Express?
- 15. © 2018 Burning Glass Technologies—Proprietary and Confidential Labor Market Entity Similarity API 15 • https://apis.burning-glass.com/demo/similarity/
- 16. © 2018 Burning Glass Technologies—Proprietary and Confidential • Customized learning recommendations based on jobseeker profile and career aspirations • BGT has the ability to tag learning content with the relevant skills taught by that content • BGT Content tagger uses a combination of multi-label classification models for occupations (~1800) and skills (~15000) 16 RECOMMEND THE LEARNING THAT DRIVES ADVANCEMENT Learning Hadoop (Lynda.com)
- 17. © 2018 Burning Glass Technologies—Proprietary and Confidential JOB DUTY KNOWLEDGE / SKILLS CAPABILITIES In Addition to Grade 11 Core Tasks, Activities & Responsibilities • Provides highest-level of crisis management requiring top technical expertise, which are vital to the success of customers and Cisco as a whole. • Identifies and works issues that impact the organization's ability to provide global world class customer support. • Contributes to or authors RFCs, internal engineering (EDCS) and TAC / FTS documents, etc. In Addition to Grade 11 Technical Knowledge & Problem Solving • Ability to configure, and troubleshoot complex multi technology / product networks. • A technical leader / SME, exhibiting exceptional talent and knowledge within the networking field. • Is widely recognized for technical accomplishments and expertise throughout TS and Services. • Has proven crisis management skills and drives change through innovation. • Go-to resource in pre-CAP and CAP situations. • Works consistently with TSPM to continue to provide thought leadership towards TS Advantage and future TS/FTS offerings. Education & Experience • MSc in Engineering, Telecommunication, or Computer Science preferred. • 10-15 years of internetworking experience required. • Should be a double CCIE. 3rd CCIE desirable. In Addition to Grade 11 Impact • Viewed as a highly skilled and expert technical resource throughout TS and across functions like Engineering, BU escalation, product supportability, testing. • Be able to author a technical publication or book in an area of expertise in the field of Routing/Switching or any other supported technology. Mentoring • Be a well-rounded person to others in the org, and as well to HTTS / TAC engineers, demonstrating best practices / work ethics in both technical abilities and written and verbal communication. • Overall, must have strong people skills, and be a role model in that respect to other HTEs. MBOs • Dedication to customer success • Communication skills • Teamwork • Business Knowledge HTE (G12)-Technical Leader II Provides advanced network level services related to reactive issues to strategic Cisco customers
- 18. © 2018 Burning Glass Technologies—Proprietary and Confidential
- 19. © 2018 Burning Glass Technologies—Proprietary and Confidential COMPENSATION INSIGHTS Role Pricing Tools: • Neural Network based salary model trained on 10M advertised salaries. • Assess specific market impact of compensable factors • Skills (from dictionary of 17,000 skills) • Location • Education • Experience • Industry Competitive Analysis • Understand market dynamics • Track competitor advertising: what jobs and skills at what price? 19
- 20. © 2018 Burning Glass Technologies—Proprietary and Confidential Job Posting Salary Prediction API 20 • Predicting Salary associated with a job posting • Understanding the impact of different variables in salary prediction • Geolocation (is NYC more expensive than Boston for the same job posting) • Title – Can we understand impact of title (level I vs Level II) in salary • Impact of higher value skills in job posting • Data Scientist1 - [Python, Java, data analysis] • Data Scientist 2- [Python, Java, ,machine learning, deep learning] • Data Analyst 1– [excel, machine learning, python] • Data Analyst2 – [word, excel, data analysis] • Return of education and experience for an occupation – Does salary changes uniformly for all occupation or differently?
- 21. © 2018 Burning Glass Technologies—Proprietary and Confidential Thank You! 21