SlideShare uma empresa Scribd logo
1 de 13
Scoring points in a Kaggle
       competition



      (lessons learned)
What was the competition about?
●   http://www.kaggle.com/c/job-salary-prediction
●   http://www.kaggle.com/c/job-salary-
    prediction/leaderboard

●   Competition about: predicting salaries from job
    postings
●   Input: ~500k postings with salary information
●   Output: 50k postings to predict
...input/train...
    {
●       "category":"Engineering Jobs",
●       "locationNormalized":"Dorking",
●       "title":"Engineering Systems Analyst",
●       "sourceName":"cv-library.co.uk",
●       "company":"Gregory Martin International",
●     "fullDescription":"engineering systems analyst dorking surrey salary ****k our client is located in dorking,
    surrey and are looking for engineering systems analyst our client provides specialist software development
    keywords mathematical modelling, risk analysis, system modelling, optimisation, miser, pioneeer engineering
    systems analyst dorking surrey salary ****k",
●       "contractTime":"permanent",
●       "locationRaw":"dorking, surrey, surrey",
●       "id":"12612628",
●       "contractType":"",
●       "salaryRaw":"20000 - 30000/annum 20-30K",
●       "salaryNormalized":25000.0
●   }
...predict...
●
    {
●
        "category":"IT Jobs",
●
        "locationNormalized":"London",
●
        "title":"lead technical architect, c banking",
●
        "sourceName":"jobserve.com",
●
        "company":"Scope AT Limited",
●
      "fullDescription":"lead technical architect required for a tier **** investment bank with excellent c skills. the main function of the role is to be the architectural lead, in
    particular designing solution architecture that will support the strategic vision. draft the roadmap for the next phase of the balance sheet management project and work with
    the business and it to then deliver this work with the business and it to design and implement the new solution to calculate the internal charge of borrowing funds within the
    group design a sophisticated liquidity reporting solution to deliver basel iii, stress testing etc. the role will focus on the following: work closely with the users, systems
    designers and the developers to design and build the required technical solution using a variety of technologies, including vendor products and inhouse built solutions
    technical design and overseer of the solution implementation for enhanced alm liquidity reporting. design and provide development oversight to all technical components
    that will exist within treasury it. design and provide technical leadership on the data acquisition, etl and storage for all common reporting requirements ensure individual
    solution designs fit within the overall strategy for treasury and all associated pillars within the program requirements: degree educated seasoned (57 years minimum)
    technical architecture experience. must demonstrate having lead technical design and/or architecture for a significant multiyear business transformational program. working
    on the design and build of a new/complex architecture with large volumes of data strong oo development background wide experience in design and build of technical
    solutions across a variety of different technologies experience working on projects that are rich in business and data complexity. technically articulate and able to
    communicate clearly to technical and treasury staff in a clear fashion ability to produce design patterns and technical framework documentation to set standards and
    patterns for the development team. c/java experience strong knowledge of investment banking functions, minimum 5 years in banking sector. strong working knowledge and
    experience in working in front to back projects; sound understanding of middle and back office functions scope at acts as an employment agency for permanent recruitment
    and employment business for the supply of temporary workers. by applying for this job you accept the t c s, privacy policy and disclaimers which can be found on our
    website.",
●       "contractTime":"permanent",
●
        "locationRaw":"London",
●
        "id":"13656201",
●
        "contractType":"",
●
        "salaryRaw":"",
●       "salaryNormalized":null
●
    }
●   It looks easy. Sort of.
●   Conceptually its easy.
●   Nothing comes for granted.
●


●   Cleaning the data: 3 days of work...
Hacking time
●   1)
         Copy paste programming. I took kaggle provided
         demo. Run it and submitted the results.
●   2)
         I have a big machine, then why not tweak a bit
         code
●   3)
         ●   First insight: clustering and ditch away the random
             forest
         ●   Implemented the clustering myself
              –   Failed
              –   Theoretical knowledge and practice are not always a happy
                  couple
Clustering problems:
●   the size of the cluster matters;
●   the salaries are sparsed for the elements in a
    cluster
●   Some terms in the documents are influcening
    the clustering
●   Decide the number of clusters
●   4) Implement the random forest myself.
    –   Fail. To much coding for selecting the features.
●   Roll back to the clustering
    –   I didn't want to write code
    –   I wanted to score points
●   Epiphany happened :D
    –   Why not use Lucene?
    –   It can provide clustering :)
The solution gets implemented
●   Transform the data into json.
●   Clean the data using stopwords.
●   Index the data in lucene.
●   Here's the cool part: MoreLikeThis query.
       ●   Start up running query
       ●   Eliminate the outliers
       ●   Done
       ●   Drawbacks:
            –   High recall
            –   Variable precision
Thanks. Questions?




●   Contact: alexandru.sisu@gmail.com
●   Twitter: twitter.com/alexsisu
●   Wanna work on cool stuff? We're hiring:)
      http://atigeo.com/Company/join.aspx

Mais conteúdo relacionado

Semelhante a Big data101kagglepresentation

DevopsBusinessCaseTemplate
DevopsBusinessCaseTemplateDevopsBusinessCaseTemplate
DevopsBusinessCaseTemplate
Peter Lamar
 
rakesh_resume_technical_latest
rakesh_resume_technical_latestrakesh_resume_technical_latest
rakesh_resume_technical_latest
paka rakesh
 

Semelhante a Big data101kagglepresentation (20)

From prototype to production - The journey of re-designing SmartUp.io
From prototype to production - The journey of re-designing SmartUp.ioFrom prototype to production - The journey of re-designing SmartUp.io
From prototype to production - The journey of re-designing SmartUp.io
 
Data engineering in 10 years.pdf
Data engineering in 10 years.pdfData engineering in 10 years.pdf
Data engineering in 10 years.pdf
 
Agile india2018 exp_report
Agile india2018 exp_reportAgile india2018 exp_report
Agile india2018 exp_report
 
Ahmed El Mawaziny CV
Ahmed El Mawaziny CVAhmed El Mawaziny CV
Ahmed El Mawaziny CV
 
How to become a data scientist
How to become a data scientist How to become a data scientist
How to become a data scientist
 
Big data and other buzzwords
Big data and other buzzwordsBig data and other buzzwords
Big data and other buzzwords
 
Maryna Sokyrko & Oleksandr Chugui: Building Product Passion: Developing AI ch...
Maryna Sokyrko & Oleksandr Chugui: Building Product Passion: Developing AI ch...Maryna Sokyrko & Oleksandr Chugui: Building Product Passion: Developing AI ch...
Maryna Sokyrko & Oleksandr Chugui: Building Product Passion: Developing AI ch...
 
Why more than half of ML models don't make it to production
Why more than half of ML models don't make it to productionWhy more than half of ML models don't make it to production
Why more than half of ML models don't make it to production
 
Business Applications of Predictive Modeling at Scale - KDD 2016 Tutorial
Business Applications of Predictive Modeling at Scale - KDD 2016 TutorialBusiness Applications of Predictive Modeling at Scale - KDD 2016 Tutorial
Business Applications of Predictive Modeling at Scale - KDD 2016 Tutorial
 
Scaling & Transforming Stitch Fix's Visibility into What Folks will love
Scaling & Transforming Stitch Fix's Visibility into What Folks will loveScaling & Transforming Stitch Fix's Visibility into What Folks will love
Scaling & Transforming Stitch Fix's Visibility into What Folks will love
 
DevopsBusinessCaseTemplate
DevopsBusinessCaseTemplateDevopsBusinessCaseTemplate
DevopsBusinessCaseTemplate
 
Machine learning in survey monkey
Machine learning in survey monkeyMachine learning in survey monkey
Machine learning in survey monkey
 
Machine learning specialist ver#4
Machine learning specialist ver#4Machine learning specialist ver#4
Machine learning specialist ver#4
 
DevOps Days Rockies MLOps
DevOps Days Rockies MLOpsDevOps Days Rockies MLOps
DevOps Days Rockies MLOps
 
Managing software projects & teams effectively
Managing software projects & teams effectivelyManaging software projects & teams effectively
Managing software projects & teams effectively
 
rakesh_resume_technical_latest
rakesh_resume_technical_latestrakesh_resume_technical_latest
rakesh_resume_technical_latest
 
"What we learned from 5 years of building a data science software that actual...
"What we learned from 5 years of building a data science software that actual..."What we learned from 5 years of building a data science software that actual...
"What we learned from 5 years of building a data science software that actual...
 
Labeling all the Things with the WDI Skill Labeler
Labeling all the Things with the WDI Skill Labeler Labeling all the Things with the WDI Skill Labeler
Labeling all the Things with the WDI Skill Labeler
 
Hats are the new leadership
Hats are the new leadershipHats are the new leadership
Hats are the new leadership
 
A Tester's Life
A Tester's LifeA Tester's Life
A Tester's Life
 

Último

Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
vu2urc
 

Último (20)

Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
Partners Life - Insurer Innovation Award 2024
Partners Life - Insurer Innovation Award 2024Partners Life - Insurer Innovation Award 2024
Partners Life - Insurer Innovation Award 2024
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law DevelopmentsTrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
 
Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a Fresher
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonets
 

Big data101kagglepresentation

  • 1. Scoring points in a Kaggle competition (lessons learned)
  • 2. What was the competition about? ● http://www.kaggle.com/c/job-salary-prediction ● http://www.kaggle.com/c/job-salary- prediction/leaderboard ● Competition about: predicting salaries from job postings ● Input: ~500k postings with salary information ● Output: 50k postings to predict
  • 3.
  • 4. ...input/train... { ● "category":"Engineering Jobs", ● "locationNormalized":"Dorking", ● "title":"Engineering Systems Analyst", ● "sourceName":"cv-library.co.uk", ● "company":"Gregory Martin International", ● "fullDescription":"engineering systems analyst dorking surrey salary ****k our client is located in dorking, surrey and are looking for engineering systems analyst our client provides specialist software development keywords mathematical modelling, risk analysis, system modelling, optimisation, miser, pioneeer engineering systems analyst dorking surrey salary ****k", ● "contractTime":"permanent", ● "locationRaw":"dorking, surrey, surrey", ● "id":"12612628", ● "contractType":"", ● "salaryRaw":"20000 - 30000/annum 20-30K", ● "salaryNormalized":25000.0 ● }
  • 5. ...predict... ● { ● "category":"IT Jobs", ● "locationNormalized":"London", ● "title":"lead technical architect, c banking", ● "sourceName":"jobserve.com", ● "company":"Scope AT Limited", ● "fullDescription":"lead technical architect required for a tier **** investment bank with excellent c skills. the main function of the role is to be the architectural lead, in particular designing solution architecture that will support the strategic vision. draft the roadmap for the next phase of the balance sheet management project and work with the business and it to then deliver this work with the business and it to design and implement the new solution to calculate the internal charge of borrowing funds within the group design a sophisticated liquidity reporting solution to deliver basel iii, stress testing etc. the role will focus on the following: work closely with the users, systems designers and the developers to design and build the required technical solution using a variety of technologies, including vendor products and inhouse built solutions technical design and overseer of the solution implementation for enhanced alm liquidity reporting. design and provide development oversight to all technical components that will exist within treasury it. design and provide technical leadership on the data acquisition, etl and storage for all common reporting requirements ensure individual solution designs fit within the overall strategy for treasury and all associated pillars within the program requirements: degree educated seasoned (57 years minimum) technical architecture experience. must demonstrate having lead technical design and/or architecture for a significant multiyear business transformational program. working on the design and build of a new/complex architecture with large volumes of data strong oo development background wide experience in design and build of technical solutions across a variety of different technologies experience working on projects that are rich in business and data complexity. technically articulate and able to communicate clearly to technical and treasury staff in a clear fashion ability to produce design patterns and technical framework documentation to set standards and patterns for the development team. c/java experience strong knowledge of investment banking functions, minimum 5 years in banking sector. strong working knowledge and experience in working in front to back projects; sound understanding of middle and back office functions scope at acts as an employment agency for permanent recruitment and employment business for the supply of temporary workers. by applying for this job you accept the t c s, privacy policy and disclaimers which can be found on our website.", ● "contractTime":"permanent", ● "locationRaw":"London", ● "id":"13656201", ● "contractType":"", ● "salaryRaw":"", ● "salaryNormalized":null ● }
  • 6. It looks easy. Sort of. ● Conceptually its easy. ● Nothing comes for granted. ● ● Cleaning the data: 3 days of work...
  • 7. Hacking time ● 1) Copy paste programming. I took kaggle provided demo. Run it and submitted the results. ● 2) I have a big machine, then why not tweak a bit code
  • 8. 3) ● First insight: clustering and ditch away the random forest ● Implemented the clustering myself – Failed – Theoretical knowledge and practice are not always a happy couple
  • 9. Clustering problems: ● the size of the cluster matters; ● the salaries are sparsed for the elements in a cluster ● Some terms in the documents are influcening the clustering ● Decide the number of clusters
  • 10. 4) Implement the random forest myself. – Fail. To much coding for selecting the features.
  • 11. Roll back to the clustering – I didn't want to write code – I wanted to score points ● Epiphany happened :D – Why not use Lucene? – It can provide clustering :)
  • 12. The solution gets implemented ● Transform the data into json. ● Clean the data using stopwords. ● Index the data in lucene. ● Here's the cool part: MoreLikeThis query. ● Start up running query ● Eliminate the outliers ● Done ● Drawbacks: – High recall – Variable precision
  • 13. Thanks. Questions? ● Contact: alexandru.sisu@gmail.com ● Twitter: twitter.com/alexsisu ● Wanna work on cool stuff? We're hiring:) http://atigeo.com/Company/join.aspx

Notas do Editor

  1. Terms – stop words can mess up the clustering? Which is the number of cluster that you need?