ITC6002 – Exploring and Analyzing Data
Institution: MSc in Data Science, The American College of Greece
Term: Fall 2023
Full Codebase and Report: GitHub Repository
Overview
This project investigates how the COVID-19 pandemic affected public trust in expert-driven forecasts, particularly epidemiological ones. By analyzing survey data on behavioral shifts, trust levels, and forecast perception during the pandemic, the team sought to understand correlations between demographic factors, media use, and belief in data-informed policy decisions.
Objectives
- Explore how trust in forecasts evolved over the course of the pandemic
- Identify demographic and behavioral variables most associated with forecast trust
- Quantify associations using regression and clustering techniques
- Visualize latent structures in trust-related survey data
Dataset
The dataset is based on a survey conducted during the COVID-19 pandemic and includes features such as:
- Age, education, country of residence
- Risk perception
- Trust in expert vs. non-expert forecasts
- Sources of COVID-related information
Methodology
- Data cleaning & wrangling with pandas
- Dimensionality reduction using PCA
- Clustering respondents via K-means and silhouette analysis
- Regression modeling for trust predictors
- Visualization using matplotlib & seaborn
Tools
- Python 3
- pandas, NumPy
- scikit-learn
- matplotlib, seaborn
- Jupyter Notebook
Key Findings
- Expert trust was higher among older and more educated respondents
- Greater reliance on social media correlated with reduced trust in expert forecasts
- PCA revealed latent dimensions tied to institutional trust and information reliability
- Cluster analysis identified three stable respondent profiles by trust orientation
Deliverables
final_notebook.ipynb: Full analysis with inline commentarycode_file.py: Modular script version of key steps- Final Report (DOCX & PDF)
- Final Presentation (PPTX)
License
For academic and demonstration purposes only.