This project focuses on the analysis of emotional patterns and their correlation with loan operations. It includes emotional pattern analysis, loan-emotion correlation, lending operation assessment, machine learning modeling, and overall visualization. The goal is to understand how emotions impact loan decisions and to provide data-driven insights into lending operations.
- Identifies dominant emotional patterns over time.
- Analyzes how emotions vary based on relationships, time of day, and other contexts.
- Examines the relationship between emotional patterns and loan terms (amount, interest rate).
- Identifies emotional factors influencing loan outcomes (approved, denied, default).
- Evaluates the effectiveness of lending policies (interest rates, loan amounts) in terms of default rates.
- Assesses loan disbursement, performance, and revenue trends over time.
- Predicts loan terms based on emotional and contextual data.
- Includes model interpretation using SHAP values.
- Visualizes emotional patterns and loan data using interactive graphs and slicers.
- Provides insights into the relationship between emotions and loan amounts, emotional intensity over time, and other key factors.
The project utilizes data from three primary sources:
- Emotional Data: Captures users' emotional states (e.g., intensity, relationship, time of day).
- Loan Data: Includes loan-related information (e.g., loan amount, interest rate, status).
- User Data: Contains user-related attributes (e.g., credit score, approved/denied dates, credit limit).
- Data Merging: Merged tables (
loans,users,emotional_data) into a single DataFrame for analysis. - Data Exploration: Conducted data exploration and identified missing values, outliers, and patterns.
- Data Preprocessing: Treated outliers, normalized the data, and handled missing values.
- Emotional Pattern Analysis: Performed visual analysis of emotional patterns and their contexts.
- Loan-Emotion Correlation: Analyzed correlations between emotional factors and loan outcomes.
- Lending Operation Assessment: Evaluated loan performance and profitability over time.
- Machine Learning: Built and evaluated predictive models for loan terms.
- Visualization: Created a comprehensive Power BI dashboard to visualize key metrics.
- Python: For data processing, analysis, and machine learning.
- Pandas: Data manipulation and cleaning.
- Power BI: For interactive data visualization and dashboard creation.
- SQLite: Database for storing and querying loan, user, and emotional data.
- Git: Version control and project management.
- Emotional Pattern Dashboard: Includes interactive bubble and bar charts displaying the relationship between emotions, loan amounts, relationships, and time of day.
- Slicers: Allow users to filter by relationship and primary emotion to explore specific patterns.
- All the required Python scripts, Jupyter Notebook, SQL query file, and Power BI dashboard file are included.
- Detailed explanations of the code, SQL query, and visualizations are provided in three separate Word documents.
- Implement advanced machine learning models for better loan term predictions.
- Add additional filtering options to the Power BI dashboard.
- Explore the integration of real-time emotional data for dynamic analysis.