Full Data Analysis Workflow on a Sales Dataset

Project Overview

This project demonstrates a complete data analysis workflow using a real-world sales dataset. The objective is to transform raw transactional data into structured insights through data cleaning, preprocessing, exploratory analysis, and visualization.

The project follows a structured analytical pipeline:

• Data inspection and merging of three source tables

• Data cleaning and validation

• Feature engineering (Revenue, Total Cost, Profit)

• Exploratory Data Analysis (EDA)

• Business insight extraction

This notebook reflects a real-world data analyst approach rather than isolated visualizations.

Tools & Libraries

• Python

• Pandas

• NumPy

• Matplotlib

• Seaborn

Dataset

The project uses three source datasets that are merged into a single analytical dataframe:

Events — transactional sales records:

• order_id – unique order identifier

• order_date – date of purchase

• ship_date – shipping date

• order_priority – order priority level

• country_code – country of sale

• product_id – product identifier

• sales_channel – sales channel (Online / Offline)

• units_sold – number of units sold

• unit_price – price per unit

• unit_cost – cost per unit

Products — product catalog with id and item_type

Countries — country reference with alpha-3, region, sub-region

Data Preparation

The following preprocessing steps were performed:

• Data structure inspection across all three tables

• Data type corrections (date conversion for order_date and ship_date)

• Missing value handling (country codes filled with "Unknown", units_sold filled with mean)

• Cyrillic-to-Latin character normalization and duplicate check

• Merging tables via product_id and country_code

• Creation of calculated metrics: Revenue, Total Cost, Profit

Exploratory Data Analysis

Revenue & Profit by Category

• The most profitable category is Cosmetics, the least profitable is Fruits.

• The highest revenue and costs are in the Office Supplies category.

• Profit distribution between Online and Offline channels is relatively close in value across categories.

Geographic Performance

• Top countries by order volume: San Marino, Andorra, Romania

• Revenue by sub-region: Southern Europe leads (33.1%), followed by Eastern Europe (22.2%) and Northern Europe (17.4%)

Shipping Interval Analysis

• Average shipping interval varies by category; the longest delays are in Office Supplies and Cereal

• By country: longest shipping time in Hungary, shortest in Croatia

• By region: delivery in Asia takes longer than in Europe

Profit vs Shipping Interval

• No direct linear relationship between shipping time and profit

• Profit is distributed evenly across the entire range (1–50 days), meaning delivery time does not significantly affect profitability

Sales Dynamics Over Time

• Revenue dynamics differ across categories and countries — no single general trend

• European region shows significantly higher values compared to Asia, with more volatile dynamics

Orders by Day of Week

• Most orders are placed on Sunday (207), fewest on Thursday (167)

• Distribution is relatively even across most categories, with some exceptions: Household sales dip on Tuesday, Fruits sales peak on weekends

Key Insights

• The most profitable category is Cosmetics; the least profitable is Fruits

• Online and Offline channels show relatively similar profit distribution across categories

• Southern Europe generates the largest share of revenue (33.1%)

• Shipping interval does not significantly affect profitability

• Delivery in Asia takes longer than in Europe

• Sunday has the highest order volume; Thursday has the lowest

• Revenue dynamics vary across categories and regions with no single universal trend

How to Run

1. Clone this repository

2. Install dependencies:

pip install -r requirements.txt

3. Open data_analysis_workflow.ipynb in Jupyter Notebook or Google Colab

Project Structure

full-data-analysis-workflow/
├── images/
│   ├── plots.png
│   ├── profit_category.png
│   ├── profit_vs_shipping.png
│   └── order_day_week.png
├── data_analysis_workflow.ipynb
├── requirements.txt
└── README.md