Research Task 05 (Part 2) – Descriptive Statistics with Grouping (Premier League Dataset)

📌 Overview

This project is Research Task 05 - Part 2, which builds upon the work completed in Research Task 05 - Part 1.
In Task 05 - Part 01, I:

• Used the Premier League 2022–23 dataset (first 10 rows only) to create descriptive statistics without grouping.
• Generated 10 natural language questions for an LLM (ChatGPT) to answer.
• Compared LLM responses with my Python-generated descriptive statistics.
• Wrote Python queries for questions that could not be directly answered from the summary.

Task 06 continues the analysis, this time grouping the dataset by stadium to generate more granular insights.

---

🏟 Task 05 - Part 2 Objective

• Perform descriptive statistics grouped by stadium.
• Create a pivot-style summary table that is more concise and easier to read than the raw grouped output.
• Limit visualizations to specific columns to focus the analysis:
  • Goals Home
  • Away Goals
  • home_possessions
  • away_possessions
  • home_chances
  • away_chances
• Draft 10 new natural language prompts for the LLM, focused on stadium-level insights.
• Write Python scripts to answer questions whose answers cannot be directly read from the grouped summary.

---

📂 Repository Contents

File Name	Description
Premier_League	Excel dataset of the 2022–23 Premier League season.
Premier_League_bar_Away_Goals.png	Bar chart of away goals by stadium.
Premier_League_bar_away_possessions.png	Bar chart of away possessions by stadium.
Premier_League_bar_Goals_Home.png	Bar chart of home goals by stadium.
Premier_League_bar_home_possessions.png	Bar chart of home possessions by stadium.
Premier_League_bar_home_chances.png	Bar chart of home chances by stadium.
Premier_League_bar_away_chances.png	Bar chart of away chances by stadium.
Prompt_log.txt	Record of LLM prompts used in Task 06.
qa_results.xlsx	Python-generated results for questions that require extra computation.
questions.txt	The 10 LLM prompts for Task 06.
summary_grouped_by_stadium_Premier_League.xls	Stadium-level descriptive statistics (pivot-style summary).
Task_06_descriptive_stats.ipynb	Jupyter Notebook containing all Python code for Task 05 - Part 2.

---

⚙️ How It Works

1. Data Loading & Cleaning

   • The dataset is read in full (unlike Task 05 where only the first 10 rows were used).
   • Missing values are handled.

2. Grouping & Summary

   • Data is grouped by stadium.
   • Summary statistics (mean, min, max, standard deviation, etc.) are computed for relevant numeric columns.
   • Output is saved as an easy-to-read pivot-style summary.

3. Visualization

   • Bar charts generated only for the six selected numeric columns.
   • Plots saved to PNG files.

4. LLM Questioning

   • 10 prompts created for ChatGPT to answer.
   • Python scripts created for non-summary questions.

---

📊 Key Learning Outcomes

This task helped me:

• Work with grouped descriptive statistics.
• Produce more readable aggregated summaries.
• Create targeted visualizations instead of plotting every numeric column.
• Compare AI-generated insights with Python-calculated results.

---

🔗 Continuation from Task 05 - Part 1

Part 2 expands upon Task 05 by introducing grouping logic and more targeted visualizations, providing deeper, stadium-specific insights.

---