This is a step by step approach to the Fraud Detection in Electricity and Gas Consumption Challenge. This is a knowledge-based challenge hosted on Zindi, and a great place to start learing practical data science.

Understanding the problem

When approaching a machine learning challenge, its important to understand the problem; this will give you a clearer picture on what the final model is expected to achieve.

You should try answering the following questions before/while building your machine learning solution:

• What question(s) am I trying to answer?
• Is my data relevant to this solution? Will my data help me answer these questions?
• How will I measure success of my machine learning model?
• How will my machine learning model help in improving my business, or will it provide insights to my research question?

It is quite possible for you as an ML practitioner to build a sophisticated and well-performing model that does not solve the right problem - by answering these questions you get rid of confusion as it gives you a blueprint on how to approach the specified problem.

Type of problem

From the description above, we know that we are building a Keep learning with Zindi by becoming a member