A Six-Step Data Science Workflow Illustrated by Food Delivery

Summary

Using a restaurant on a food delivery app as an example, the document presents a general six-step data science workflow: define the business problem, collect data, clean and remediate it, analyze and interpret it, model it, and communicate findings. The restaurant’s goal is to improve customer service and ratings. Suggested questions include delivery times, popular menu items, order locations, and customer reviews. The article distinguishes primary data collection from reuse of existing data and stresses that missing or inconsistent records can undermine later analysis.

For analysis, it recommends exploring relevant variables and visualizing comparisons such as current and prior-year sales. It then describes organizing calculations into a model, using statistical or machine-learning methods as appropriate, and presenting results in reports or dashboards for business decisions. A small example reports an average delivery time of 38.1 for ten recent orders. This is an introductory framework rather than a detailed technical tutorial: it gives no model specification, validation approach, or evidence that the proposed business changes improve outcomes. Its examples concern food delivery, not trading.

Key ideas

• Start with a clearly framed business question before gathering data.
• Data may be collected directly or reused from existing sources.
• Clean missing, inconsistent, or misspelled data before analysis.
• Use exploratory analysis and visualizations to understand variables and patterns.
• Choose statistical or machine-learning models to fit the problem, then communicate findings clearly.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.