Comparative Analysis of Financial Datasets Predictive Modeling Using Feature Extraction Methods

In the field of finance, predictive modeling is crucial for analyzing and forecasting financial trends and their effects. The feature extraction stage of the predictive modeling method is crucial and involves selecting and identifying relevant variables or features from financial data. To determine which feature extraction methods are most successful at locating the most illuminating and foretelling qualities, it is critical to evaluate and contrast various feature extraction strategies. Furthermore, it might be difficult to assess the efficacy and correctness of such a large amount of data, such as the reports from the Management Data and Analysis (MD&A) dataset. We used CountVectorizer, TF-IDF, sentiment, Principal Component Analysis (PCA), Recursive Feature Elimination (RFE), Lasso Regularization, and Random Forest Feature along with three well-known machine learning models, namely Logistic Regression, Support Vector Machine (SVM), and K-Nearest Neighbor (KNN), to address these challenges. Finally, utilizing various evaluation matrices like precision, recall, and F1 Score, we evaluated and contrasted their output.