CS3121 - Online Purchase Intentions During Crises

Group Details

• Project Group Name: Project Group 13

Project Overview

In this project, we analyze consumer behavior during crises in Sri Lanka, focusing on their intentions to make online purchases. Conducted on behalf of Wolt, a leading retail operator, our study utilizes 836 survey responses to examine how various psychological, social, and system-related factors influence online shopping decisions during crisis periods. Key components include data preprocessing, statistical analysis, hypothesis testing, and prescriptive rule mining using the Apriori algorithm.

---

Folder Structure

/OnlinePurchaseIntention/
│
├── README.md                         ← Project overview and instructions
├── requirements.txt                  ← Python package dependencies
│
├── data/
│   └── responses.xlsx                ← Raw survey data (836 responses)
│
├── preprocessing/
│   └── data_cleaning.ipynb          ← Missing values, duplicates, transformations, reliability checks
│
├── da_eda/
│   └── descriptive_analysis_and_EDA.ipynb  ← Descriptive statistics, visualizations, trend analysis
│
├── hypothesis_testing/
│   ├── given_hypotheses.ipynb       ← Hypothesis tests as per project brief
│   └── custom_hypotheses.ipynb      ← Additional hypotheses based on conceptual model
│
├── rule_mining/
│   └── apriori_rules.ipynb          ← Association rule mining using Apriori algorithm

---

Setup Instructions

1. Clone the Repository

git clone https://github.com/InduwaraRathnayake/CS3121-OnlinePurchaseIntention.git

2. Set Up Virtual Environment or Use Google Colab

python3 -m venv venv
source venv/scripts/activate 

3. Install Dependencies

pip install -r requirements.txt

4. Run Notebooks

• Navigate to the relevant folders (preprocessing, eda, etc.)
• Open Jupyter Notebook or VS Code
• Run the .ipynb files

---

Key Tasks

Task	Description
1. Data Preprocessing	Cleaned dataset, handled missing values, calculated Cronbach's alpha for reliability
2. EDA	Explored variable distributions, trends, outliers, and relationships using statistical and visual tools
3. Hypothesis Testing	Tested 10 given and 10 additional hypotheses using regression, correlation, mediation analysis
4. Rule Mining	Applied Apriori algorithm to extract 5 actionable rules from consumer behavior patterns

---

References

• Pandas Documentation: https://pandas.pydata.org/docs/
• Cronbach's Alpha: https://en.wikipedia.org/wiki/Cronbach%27s_alpha
• Apriori Algorithm: https://en.wikipedia.org/wiki/Apriori_algorithm