Well, I don’t want to go over another post on Data Science framework that you can find on the Internet by just googling it. I like the CRISP-DM life cycle and widely used and know in the industry. You can find the details here. However, I see a gap in educational content over the framework widely used in Industry.

Below are the steps of the newly adopted framework. In essence, it’s an extension of CRISP-DM. It is the sequence of the phases and is not strict and moving back and forth between different stages as it is always required.

‘Data Science Framework in Industry

Below is the overview of DS framework, I will use this DS framework and solve a business problem in future post:

  1. Business Requirement

     - Define Objectives

 - Define outcomes (feedback)

  2. Data Acquisition

     - Collect Data

 - Consult/work with/as Data Engineers

  3. Data Preparation

     - Understand and Highlight problem areas, process, and caveats

  4. Exploration of Data Analysis

     - Slice and Dice Data,

 - Plots Understand Data edge cases

  5. Modelling , Evaluation & Interpretation

     - Develop the algorithm to solve a problem

 - Highlight different solution and recommendation

  6. Communicate Results

     - Capture outcome after implementation

 - Consult with business stakeholders to share the result

  7. Deployment strategy

     - Deployment (online / offline), API integration, Schedule..etc

 - Consult / work with Data Engineers /  Software Engineers for integration / Production

  8. Testing framework

     - Measure Result by A/B testing

 - Consult/work with stakeholders to share the result

  9. Business Approval

     - Showcase the **financial impact **

  10. Operations

     - Create Standard Reporting framework to measure/track the performance of algorithms

 - Retrain models, handle failures, a process built

 - Consult/work with stakeholders to share the result

  11. Optimization

    - Improve models

- more data

- more features code improvement

- process improvement

Reference:

✠  Previous Beware of #fakeguru - AI/ML/DS ✠  Next Customer Churn