Developed Multinomial Logistic Regression for Voter Analysis

• Day: 2024-09-09
• Time: 15:20 to 16:45
• Project: Teaching
• Workspace: WP 1: Strategic / Growth & Development
• Status: Completed
• Priority: MEDIUM
• Assignee: Matías Nehuen Iglesias
• Tags: Multinomial Logistic Regression, Voter Behavior, Data Analysis, Education, Python

Description

Session Goal

The session aimed to explore and implement multinomial logistic regression models to analyze voter behavior, particularly focusing on transition probability models and voting patterns.

Key Activities

• Framework Design: Established a framework for modeling voter behavior using transition probability models, focusing on multinomial logistic regression and Markov processes.
• Question Design: Created a challenging question template to test understanding of multinomial logistic regression.
• Package Installation: Installed [[python]]-Levenshtein to enhance the performance of the fuzzywuzzy package.
• Model Implementation: Developed a methodology for modeling voting probability using multinomial logistic regression with Python’s statsmodels library.
• Data Cleaning: Addressed data type errors in the regression model to ensure numeric compatibility.
• Educational Insight: Provided a structured approach for students to reconstruct regression models using group sizes from contingency tables.

Achievements

• Successfully implemented multinomial logistic regression models to analyze voter behavior among Physics and Biology students.
• Enhanced understanding of statistical modeling through educational templates and problem statements.

Pending Tasks

• Further validation of models with additional datasets from different demographics.
• Exploration of alternative statistical models for comparison.

Evidence

• source_file=2024-09-09.sessions.jsonl, line_number=0, event_count=0, session_id=b1ed2efa10a4c11f27a4fb97b6838586ae8e03907ce487a93de1037fad8e1ad6
• event_ids: []