big_four

Prediciting Public Company Financial Statement Auditor

Using supervised learning classification techniques, created model that predicts if a company will be audited by the Big Four or another accounting firm

---

Target Variable and Features

Target Variable: 1)Big Four, or 2)Other

Features:

1. Cash Flows (feature engineered to indicate inflow/outflow)
   • CAPEX
   • NCF - Business Acquisitions and Disposals
   • NCF - Cash & Cash Equivalents
   • NCF - Financing
   • Issuance/Repayments of Debt Securities
   • Issuance/Purchase of Equity Shares
   • Payment of Dividends & Other Cash Distributions
   • NCF - Investing
   • NCF- Operations
2. Balance Sheet
   • Assets
   • Cash and Cash Equivalents
   • Liabilities
   • Debt
   • Equity
3. Earnings
   • Revenue
   • EBITDA
   • Equity
4. Market
   • Market Capitilization
   • Enterprise Value
5. Industry
6. Location(State)
7. Exchange

---

Data Used

PCAOB Auditor Search: https://pcaobus.org/resources/auditorsearch

Quandl - Core US Fundamentals Data: https://www.quandl.com/databases/SF1/data

---

Tools and Packages

1. Python
2. PostgreSQL
3. Pandas
4. NumPy
5. matplotlip
6. seaborn
7. Geopandas
8. scikit-learn
9. SQLAlchemy

---

Repo Guide

Database Exploration, Creation, and Querying

• explore_csv.ipynb
• create_database.sql
• query.sql

Preprocessing and EDA

• import_preprocess.ipynb
• eda_viz.ipynb

Modeling and Metrics

• knn.ipynb
• logistic_regression.ipynb
• naive_bayes.ipynb
• decision_tree.ipynb
• random_forest.ipynb
• xgboost.ipynb
• ensemble.ipynb
• roc_curve.ipynb