Advanced Transaction Analytics and Customer Segmentation for Optimized Sales Strategies

Data Loading and Initial Exploration Libraries

Loaded Dataset: I began by loading the dataset from a MySQL database using SQLAlchemy. After establishing the connection to the database, I retrieved the necessary data into a pandas DataFrame , and into a CSV file for further analysis.

Imported: I imported essential libraries including numpy, pandas, seaborn, matplotlib, and plotly for data manipulation, visualization, and analysis.

Data Import: I loaded the dataset using pd.read_csv() from the data.csv file, which includes sales transaction details.

Basic Inspection: I used functions like df.head(10), df.describe().T, and a custom summary() function to inspect the dataset’s structure, identify missing values, and compute basic statistics.

Data Cleaning and Feature Engineering

Handling Missing Values: I dropped rows with missing values in critical columns such as CustomerID and Description. Transaction Status: I created a new feature, Transaction_Status, to categorize transactions as ‘Cancelled’ or ‘Completed’ based on the InvoiceNo.

Correcting Stock Codes: I filtered out incorrect or anomalous stock codes by identifying codes with too few digits. Removing Negative Prices: I removed rows with non-positive UnitPrice to ensure accuracy in subsequent analyses.

Exploratory Data Analysis

(EDA) Top Products and Sales: I identified and visualized top-selling products and total sales by country using Plotly to create interactive charts.

Customer Behavior: I performed RFM (Recency, Frequency, Monetary) analysis to segment customers based on their purchasing behavior.

Product Recommendations: I developed a recommendation system based on customer segmentation to suggest products to customers.

Customer Segmentation and Clustering

RFM Analysis: I segmented customers using RFM metrics to understand their purchasing behavior.

K-Means Clustering: I applied K-Means clustering to group customers into distinct segments, determining the optimal number of clusters using silhouette scores and the elbow method.

Customer Profiling: I analyzed clusters to identify behavior patterns and highlighted the top products for each segment.

Time-Series Analysis Sales Trends: I conducted a time-series analysis of daily sales using seasonal decomposition to identify trends, seasonality, and residuals.

Final Insights

Pareto Analysis: I applied the Pareto principle to identify the small percentage of customers contributing to the majority of sales.

RFM Merging: I merged the results of RFM analysis with customer purchase data to create a comprehensive view of customer segments.

Visualization and Reporting: I created various visualizations to summarize key insights, including bar charts for sales by country, daily sales trends, and customer segmentation.