Planning Your Data Science Projects

Planning Your Data Science Projects

• 1.
  Spotle.ai Study Material Spotle.ai/Learn Planning SuccessfulData Science Projects
• 2.
  Spotle.ai Study Material Spotle.ai/Learn2 Project Management is the practice of initiating, planning, executing, controlling, and closing the work of a team to achieve specific goals and meet specific success criteria at the specified time.  - Wikipedia What Is Project Management? Spotle.ai Study Material Spotle.ai/Learn
• 3.
  Spotle.ai Study Material Spotle.ai/Learn ✓Strategic Alignment – Aligns all stakeholders on a common goal and roadmap ✓ Manage Budget – Ensures a work is delivered within budget ✓ Manage Time – Ensures a work is delivered within budget ✓ Manage Scope – Control to prevent scope creep ✓ Risk Management – ✓ Reuse Success Stories – Codified best practices that can be successfully reused ✓ Builds a clear Go-To-Market path – Creates a unified plan from requirement gathering to live deployment and user acceptance Why Is It Important? Spotle.ai Study Material Spotle.ai/Learn
• 4.
  Spotle.ai Study Material Spotle.ai/Learn ProjectManagement basically gives a way of managing chaos and giving a predictable path to success. 4
• 5.
  Spotle.ai Study Material Spotle.ai/Learn 5 InData Science Projects, Project Management Becomes Critical: ๏ Large In Scale ๏ Cross Functional Teams ๏ CXO Involvement ๏ High Priority ๏ High Risk ๏ Multi-Agency ๏ High Complexity ๏ Data Issues ๏ Iterative Process
• 6.
  Spotle.ai Study Material Spotle.ai/Learn 6 85%Of Big Data Projects Fail Source: eweek.com
• 7.
  Spotle.ai Study Material Spotle.ai/Learn 7 DeliveringA Successful Data Science Project Requires Smart Planning And Execution.
• 8.
  Spotle.ai Study Material Spotle.ai/Learn 8 The Methodology For Managing Data Science Projects SMART
• 9.
  Spotle.ai Study Material Spotle.ai/Learn Problem Definition •Understand the business problem • Review literature • Background Research Data Acquisition And Preparation • Identify Data Sources • Define Data Collection Plan • Data Collection – Primary and Secondary Data • Data cleaning and preparation Model Selection And Fitment • Compare and select model Validation • Validate model with actual (mock/ scrambled) data • User Acceptance Testing Deployment • Deployment of model in live Monitoring • Monitor accuracy in live scenario The Spotle.ai 6-Phase SMART Model For Data Science Projects. This model is based on industry guidelines such as CRISP-DM and the experience of practitioners. Specify and Prepare Model Analyse Roll-out Test and Monitor Feedback 9 SMART
• 10.
  Spotle.ai Study Material Spotle.ai/Learn Problem Definition Data AcquisitionAnd Preparation Model Selection And Fitment Validation Deployment Monitoring 10 Indicative Tools Required At Each Phase • Microsoft Project • Visio • Word • Powerpoint • JIRA/ Other Project Management Tool • Tableau • Power BI • Excel • Ggplot using R • Scala/ Python/ Tensorflow • Hadoop • MySQL/ Other Database • Statistics • R • Python • SQL • R • Python • Scala • Tensor Flow • SQL • Azure ML/ IBM Blue mix • Hadoop • R/ Python/ Scala • Server Tools • Monitoring and logging tools • JIRA to record feedback SMART
• 11.
  Spotle.ai Study Material Spotle.ai/Learn Problem Definition Data AcquisitionAnd Preparation Model Selection And Fitment Validation Deployment Monitoring 11 Key Stakeholders At Each Stage Of A Data Science Project • Business User • CXO/ Project Sponsor • Business Analyst • Data Scientist • Data Analysts • Data Collection Agencies • Analyst Relations • Data Scientist • Data Scientist • Software Engineering Team • Data Scientist • Business Analyst • Business User • Quality Assurance • Data Scientist • Software Engineer • Business User • Project Sponsor/ CXO • Data Scientist SMART
• 12.
  Spotle.ai Study Material Spotle.ai/Learn TheCRISP Methodology For Managing Data Science Projects Spotle.ai Study Material Spotle.ai/Learn
• 13.
  Spotle.ai Study Material Spotle.ai/Learn 13 CRISP-DMstands for cross-industry process for data mining. The CRISP- DM methodology is a structured approach to planning a data mining/ data science/ analytics project.
• 14.
  Spotle.ai Study Material Spotle.ai/Learn BusinessUnderstanding Data Understanding Data Preparation Modelling Evaluation Deployment The CRISP Model 14
• 15.
  Spotle.ai Study Material Spotle.ai/Learn Why Data Science Projects Fail? Lackof access to data Data Issues Lack of cross- functional expertise Competency Issues Complex models/ Under or Over- Fitment Other Issues Faulty data/Data misclassific ation Data Scientist lacks experience Lack of firm support from CXOs/ sponsors 15
• 16.
  Spotle.ai Study Material Spotle.ai/Learn LearnData Science In The Most Intense Course Ever: https://spotle.ai/learn/data-science 1 Manage Cross- Functional Team 6 Effectively Manage Scope, Budget And Time 4 Focus On Measurable Results 2 Engage Effectively With Senior Leaders 3 Develop Domain And Technology Understanding 5 Tolerate Uncertainty To Be A Successful Data Science Project Manager, You Should Be Able To