9 Steps to Build your first Machine Learning Use Case

1. Data Analysis from Domain Perspective - Explore your data and understand the entities, transactions, business flow captured

2. Data Science Use Cases Review from Customer / Sales / Business Opportunities perspective - Read / Connect with Stakeholders find the use cases that business needs

3. Mapping of Available Data to Data Science use cases - Conduct the feasibility study to map the use cases and data flow captured to explore about them

4. Pick and chose use case based on Data, Impact to business - With #1 and #2 find the sweet spot to hit your first use case

5. Develop Model, Build Features, Test and Validate Accuracy 

6. Demonstrate the model, explain the data, features - Demo it to business users, Sell your use case

7. 70-80% Accuracy is Good Enough to get started - Quality improves over time and more features captured. Layout plan to add more features

8. Build End to End Data Pipeline, Model Training Pipeline, Deploy as API / Expose the output as Report / API response

9. Include the aspects of Model Re-training to keep up with Changing Data Dynamics, Refresh the model over periodic intervals

The Journey Continues!!!

要查看或添加评论,请登录

Sivaram A.的更多文章

社区洞察