Financial Performance Analysis

📋 Project Overview

This project conducts a comprehensive analysis of companies' financial data to identify patterns and trends that affect EBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization). The analysis aims to provide actionable insights for data-driven business decision-making.

🎯 Main Research Question

Does financial performance affect the EBITDA of a company?

🔍 Sub-questions Addressed

• What is the relationship between Enterprise Value and EBITDA?
• What is the relationship between Market Capitalization and EBITDA?
• What is the relationship between Revenue Growth and EBITDA?
• What is the relationship between Net Income Growth and EBITDA?
• What is the relationship between Debt Growth and EBITDA?

📊 Dataset Description

The dataset contains financial statistics for 1,577 companies including:

• Financial ratios
• Growth rates
• M-Score values
• EBITDA values
• Gross profit and net profit margins
• Return on assets
• Enterprise value and market capitalization data

Dependent Variable: EBITDA
Independent Variables: Enterprise Value, Market Capitalization, Revenue Growth, Net Income Growth, Debt Growth

🛠 Methodology & Tools

Analysis Techniques

1. Descriptive Analysis

   • Measures of Central Tendency (Mean, Median, Mode)
   • Measures of Dispersion (Range, Variance, Standard Deviation, IQR)

2. Regression Analysis

   • Linear regression models for each independent variable
   • Correlation coefficient analysis
   • R-squared evaluation

3. Data Visualization

   • Excel dashboards with interactive charts
   • Tableau dashboards for advanced visualization

4. Predictive Analytics

   • Machine Learning models using Python
   • SMOTE (Synthetic Minority Oversampling Technique)
   • Fraud detection using M-Score values

Tools & Technologies

• Microsoft Excel - Data cleaning, descriptive analysis, dashboards
• Tableau - Advanced data visualization and interactive dashboards
• Python - Machine learning and predictive analysis
• Google Colab - Cloud-based Python environment
• Libraries Used: Scikit-learn, Pandas, NumPy, Matplotlib

📈 Key Findings

Descriptive Statistics (EBITDA)

• Mean: $388,207,231.7
• Median: $211,556,000
• Mode: $114,044,000
• Distribution: Positively skewed (Mean > Median > Mode)
• Range: $4,772,791,000
• Standard Deviation: $589,270,896

Regression Analysis Results

Variables	R² (Excel)	R² (ML)	Correlation (r)	Relationship Strength
EBITDA vs Enterprise Value	0.410	0.390	0.640	Strong
EBITDA vs Market Cap	0.270	-0.003	0.520	Moderate
EBITDA vs Revenue Growth	0.003	-0.003	0.060	Weak
EBITDA vs Net Income Growth	0.005	-0.003	0.070	Weak
EBITDA vs Debt Growth	0.002	-0.002	0.050	Weak

Top Performing Companies

By Revenue Growth:

• EC: 0.8%
• CLVT-PA: 0.4%
• TSM: 0.3%
• SONY: 0.1%

📊 Visualizations

The project includes:

• Excel Dashboard: Pie charts showing EBITDA distribution across companies, line graphs comparing accounts payable vs receivables
• Tableau Dashboard: Interactive visualizations of revenue growth, gross profit margins, and accounts receivables analysis

🔬 Machine Learning Models

Models Implemented

1. Linear Regression - Primary analysis tool
2. Logistic Regression - Classification accuracy testing
3. SMOTE Balancing - Handling imbalanced datasets
4. Fraud Detection - Using M-Score values

Model Performance

• Enterprise Value vs EBITDA: 39% accuracy (R² = 0.39)
• Revenue Growth predictions: 83.54% accuracy (Logistic Regression)
• Fraud Detection: 95.25% prediction score

⚠️ Limitations

• Dataset limitations and potential observer bias
• Removal of outliers using IQR may have reduced dataset size
• Machine learning models show varying accuracy levels
• Some null values and data quality issues affect predictive accuracy

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.