Processed Financial Behavior Dataset for Neuro-Fuzzy Modeling (ANFIS) Based on Investment Survey Data

Description

This dataset is derived from a large-scale online investment questionnaire conducted in 2019 and published on the Hungarian financial portal “Portfolio.hu”, developed in collaboration with Corvinus University of Budapest and Dorsum Ltd. The dataset consists of 1,542 individual responses and focuses exclusively on financial awareness and risk appetite (Page 2 of the questionnaire). The raw survey responses were transformed through a structured data preprocessing pipeline, including categorical encoding, binary expansion of multi-choice variables, missing value removal, and Min–Max normalization. The final dataset represents a structured financial behavior feature space used for clustering and neuro-fuzzy modeling (ANFIS). It captures cognitive and behavioral dimensions of investor decision-making, including risk perception, investment preferences, and financial literacy. This dataset has been used in multiple studies by the authors; however, the version provided here corresponds specifically to the preprocessing pipeline used in the associated ANFIS-based research framework. Detailed variable definitions, coding schemes, and preprocessing steps are provided in the accompanying README file. This dataset contains survey-based financial behavior data used to analyze individuals’ financial awareness, investment preferences, and risk appetite. The data were collected through a structured questionnaire and then transformed into numerical format based on a predefined coding scheme. The dataset includes three main components: (1) coded questionnaire responses, (2) normalized input data prepared for Adaptive Neuro-Fuzzy Inference System (ANFIS) modeling, and (3) a final clustered dataset with assigned risk levels. The final dataset integrates multiple variables representing financial attitudes, investment behavior, and decision-making patterns. Clustering techniques were applied to segment individuals into distinct groups based on behavioral similarities. Each observation is assigned to a cluster and labeled with a corresponding risk level category (Low Risk, Medium Risk, or High Risk). These classifications are derived from patterns in financial behavior and investment-related responses. The dataset is suitable for research in financial behavior analysis, risk profiling, machine learning applications, clustering analysis, and intelligent decision-support systems such as ANFIS.

Steps to reproduce

-Step 1: Data Acquisition and Initialization -Step 2: Data Cleaning and Preprocessing -Step 3: Multi-Answer Encoding -Step 4: Data Normalization -Step 5: Unsupervised Clustering for Label Generation -Step 6: Feature Selection for ANFIS -Step 7: Dataset Split -Step 8: ANFIS Model Construction -Step 9: Rule Base Initialization -Step 10: Model Training -Step 11: Model Evaluation -Step 12: Rule Extraction -Step 13: Queen Bee Optimization (QBE) -Step 14: Rule Pruning (Sparsification) -Step 15: Model Comparison -Step 16: Visualization -Step 17: Final Model Deployment

Institutions

• Corvinus University of Budapest

  Budapest, Budapest

Categories

Licence