Se ha denunciado esta presentación.
Utilizamos tu perfil de LinkedIn y tus datos de actividad para personalizar los anuncios y mostrarte publicidad más relevante. Puedes cambiar tus preferencias de publicidad en cualquier momento.

moOn: A Multidimensional Graph Approach to Human Resources Analytics, aizoOn

505 visualizaciones

Publicado el

GraphConnect Europe 2017
Claudio Borile, aizoOn

Publicado en: Tecnología
  • Sé el primero en comentar

moOn: A Multidimensional Graph Approach to Human Resources Analytics, aizoOn

  1. 1. moOn A Multidimensional Graph Approach to Human Resources Analytics CLAUDIO BORILE GRAPHCONNECT MAY 2017 – QEII CENTER LONDON
  2. 2. aizoOn | Ver. 1.2 11/05/2017 - Claudio Borile GraphConnect London 2 WHO WE ARE Our vision: To apply widely the scientific and quantitative approach, for a more responsible and sustainable society Our mission: To sustain the future of our customers in the digital era, bringing knowledge in technology and innovation AIZOON IS aizoOn is a technology consulting agency for innovation, is independent, and operates at a global scale
  3. 3. aizoOn | Ver. 1.2 WHO WE ARE 11/05/2017 - Claudio Borile GraphConnect London 3 We follow our customers in all continents: Africa, America, Asia, Europa, Oceania GLOBAL FOOTPRINT AUSTRALIA Sydney NSW EUROPE Torino ITA | Cuneo ITA | Milano ITA | Genova ITA Bologna ITA | Roma ITA | Bari ITA | Sheffield UK USA New York NY | Troy MI | Cambridge MA | Lewiston ME aizoOn USA aizoOn AU aizoOn EU Direct presence Areas of intervention
  4. 4. aizoOn | Ver. 1.2 “Standard” HR 11/05/2017 - Claudio Borile GraphConnect London 4 Historically the organization is resumed in a pyramidal chart, from the CEO/manager to the working base and neatly divided into departments, groups, hierarchical chain, etc.
  5. 5. aizoOn | Ver. 1.2 11/05/2017 - Claudio Borile GraphConnect London 5 “Standard” HR Reality is different, and messier
  6. 6. aizoOn | Ver. 1.2 11/05/2017 - Claudio Borile GraphConnect London 6 “Standard” HR A company is a complex organism, composed of many different interacting components.
  7. 7. aizoOn | Ver. 1.2 Data-driven HR 11/05/2017 - Claudio Borile GraphConnect London 7 In the last few years data analytics and quantitative methods have been applied to many areas of business, marketing and production to help making better choices. People are rightfully considered the most important assent in an organization, and why shouldn’t we exploit data to better know this asset?
  8. 8. aizoOn | Ver. 1.2 Data-driven HR 11/05/2017 - Claudio Borile GraphConnect London 8 In the last few years data analytics and quantitative methods have been applied to many areas of business, marketing and production to help make better choices. People are rightfully considered the most important assent in an organization, and why shouldn’t we exploit data to better know this asset?
  9. 9. aizoOn | Ver. 1.2 People Analytics 11/05/2017 - Claudio Borile GraphConnect London 9 People Analytics is a data-driven approach to the management of the workplace for a better knowledge of the real organizational structure, practices and processes. • Organization • People Review • HR Transformation • Talent Management • Perception vs. Reality • Monitoring • Early detection • Top and weak performers
  10. 10. aizoOn | Ver. 1.2 11/05/2017 - Claudio Borile GraphConnect London 10 The moOn approach moOn: MultidimensiOnal cOmpany Navigator is a People Analytics tool imagined not to substitute, but to support the traditional HR methodologies with the help of quantitative measures. Multidimensional: We use diversified sources of data to highlight different aspects of the organization and the people in it, and a coherent framework allows to navigate through these many layers of information.
  11. 11. aizoOn | Ver. 1.2 11/05/2017 - Claudio Borile GraphConnect London 11 Data Data will come from: • E-mails • CRM, Organizational and personal registries • Surveys • Meetings • Resume • … The project is now TRL 6
  12. 12. aizoOn | Ver. 1.2 11/05/2017 - Claudio Borile GraphConnect London 12 Data In this presentation: • E-mails • CRM, Organizational and personal registries • Surveys • Meetings • Resume • … The project is now TRL 6
  13. 13. aizoOn | Ver. 1.218/05/2017 13 Components Network Analysis Machine learning Techniques Dashboard & control panel Network visualization and exploration Storage Analysis Data Viz
  14. 14. aizoOn | Ver. 1.2 Mail Network – raw data 11/05/2017 - Claudio Borile GraphConnect London 14 One year of e-mail logs from a standard Microsoft exchange mail server ~ 2 million rows csv file Data format: Data is anonymized for privacy reasons timestamp sender recipient(s) Subject
  15. 15. aizoOn | Ver. 1.2 Mail Network – Graph construction 11/05/2017 - Claudio Borile GraphConnect London 15 We keep only “relevant” communications (~30% of total) Atomic resolution for internal addresses, domain level for externals A directed and weighted graph is extracted where nodes are internal people or external domains and edges represents single mail threads
  16. 16. aizoOn | Ver. 1.2 Database – E-mails 11/05/2017 - Claudio Borile GraphConnect London 16 The Neo4J graphDB stores all the information from the preprocessed dataset. The schema allows to easily and rapidly gather all time- dependent, aggregate data for later visualization, and the navigable graphs, with standard queries.
  17. 17. aizoOn | Ver. 1.2 Database – E-mails 11/05/2017 - Claudio Borile GraphConnect London 17 All the queries to the DB are done directly from moOn’s Python core using Cypher for later manipulation
  18. 18. aizoOn | Ver. 1.2 Mail Network – Overview 11/05/2017 - Claudio Borile GraphConnect London 18 ~1200 Vertices 500 internal nodes, 700 external domains ~16000 Edges
  19. 19. aizoOn | Ver. 1.2 Mail Network – Departments 11/05/2017 - Claudio Borile GraphConnect London 19 Inter-departments communications and silos
  20. 20. aizoOn | Ver. 1.2 Mail Network – People descriptive 11/05/2017 - Claudio Borile GraphConnect London 20 People workload and total contacts
  21. 21. aizoOn | Ver. 1.2 Mail Network – People descriptive 11/05/2017 - Claudio Borile GraphConnect London 21 People workload and total contacts
  22. 22. aizoOn | Ver. 1.2 Mail Network – People descriptive 11/05/2017 - Claudio Borile GraphConnect London 22 People workload and total contacts
  23. 23. aizoOn | Ver. 1.2 Mail Network – Socialization 11/05/2017 - Claudio Borile GraphConnect London 23 Socialization process for newly hired people
  24. 24. aizoOn | Ver. 1.2 Mail Network – Socialization 11/05/2017 - Claudio Borile GraphConnect London 24 Socialization process for newly hired people
  25. 25. aizoOn | Ver. 1.2 Mail Network – Stress 11/05/2017 - Claudio Borile GraphConnect London 25 We can easily monitor stress components like working after office hours, or working during weekends. Also, we can have suggestions on the daily routine of the person
  26. 26. aizoOn | Ver. 1.2 Mail Network – Stress 11/05/2017 - Claudio Borile GraphConnect London 26 We can easily monitor stress components like working after office hours, or working during weekends. Also, we can have suggestions on the daily routine of the person
  27. 27. aizoOn | Ver. 1.2 Mail Network – Communication network 11/05/2017 - Claudio Borile GraphConnect London 27 Subdivision of contacts by single addresses, departments, external domains.
  28. 28. aizoOn | Ver. 1.2 Mail Network – Communication network 11/05/2017 - Claudio Borile GraphConnect London 28 Subdivision of contacts by single addresses, departments, external domains.
  29. 29. aizoOn | Ver. 1.2 Mail Network – Communication network 11/05/2017 - Claudio Borile GraphConnect London 29 Subdivision of contacts by single addresses, departments, external domains.
  30. 30. aizoOn | Ver. 1.2 Mail network – Graph exploration 11/05/2017 - Claudio Borile GraphConnect London 30 Complete network with internals and externals separation
  31. 31. aizoOn | Ver. 1.2 Mail network – Graph exploration 11/05/2017 - Claudio Borile GraphConnect London 31 Egocentric network of a specific user
  32. 32. aizoOn | Ver. 1.2 Mail network – Graph exploration 11/05/2017 - Claudio Borile GraphConnect London 32 Departments’ network
  33. 33. aizoOn | Ver. 1.2 Mail network – Graph exploration 11/05/2017 - Claudio Borile GraphConnect London 33 Intergroup bridging and gateways to the exterior. Individual or departmental level of “frontier” towards the exterior
  34. 34. aizoOn | Ver. 1.2 Surveys 11/05/2017 - Claudio Borile GraphConnect London 34 Surveys allow us to add a layer of informal and personal network of connections between people in the organization. Easily compiled and submitted through a ad hoc web platform, surveys are automatically integrated in moOn and elaborated.
  35. 35. aizoOn | Ver. 1.2 Surveys – Graph exploration 11/05/2017 - Claudio Borile GraphConnect London 35 All methods and graphic interfaces are similar to the mail part, but they carry very different information
  36. 36. aizoOn | Ver. 1.2 Multidimensional navigation 11/05/2017 - Claudio Borile GraphConnect London 36 We look at the same people from different perspectives, to capture all the complexity of the workplace
  37. 37. aizoOn | Ver. 1.2 Network Metrics 11/05/2017 - Claudio Borile GraphConnect London 37 The topological structure of the networks yield relevant information, that we translate in easy-to-interpret visualizations
  38. 38. aizoOn | Ver. 1.2 Network Metrics 11/05/2017 - Claudio Borile GraphConnect London 38 Depending on which level of network we are focusing on, we can exploit its topological structure to infer information on the behavior and characteristics of people. Reference figures (e.g. innovators, mentors, etc.), top performers, bridges, or the frontier and bulk people with respect to the exterior of the organization or of a department. We can see the real interaction and composition of working groups, or how to easily reach a customer. For the succession problem we can compare the network structure of similar workers, compare it with other data like skills, seniority, contractual level, etc. to have the best replacement.
  39. 39. aizoOn | Ver. 1.2 Conclusions 11/05/2017 - Claudio Borile GraphConnect London 39 • Simple sources of information and data can help us know and manage our organization better. • Multidimensionality gives us the opportunity to scrutinize our organization on various levels, from the formal to the informal, and compare them to the official structures and hierarchies. • Fine and coarse grained analytics give us ready-to-use information from single employees to the whole business. • The specific choice of the Neo4j graphDB helps us to organize and query the great amount of data that we obtain for later processing.
  40. 40. www.aizoongroup.com claudio.borile@aizoongroup.com Claudio Borile @aizoongroup AUSTRALIA Sydney NSW EUROPE Torino ITA | Cuneo ITA | Milano ITA | Genova ITA Bologna ITA | Roma ITA | Bari ITA | Sheffield UK USA New York NY | Troy MI Cambridge MA | Lewiston ME THANK YOU!

×