Capstone Project: AI Study Assistant

Congratulations! You have learned Python from scratch, mastered data structures, built ML models, connected to AI APIs, and even built a RAG system. Now it is time to combine everything into one complete project.

This project will be a portfolio piece — something you can show to employers to prove your AI & ML skills.

---

1. What We Will Build

An AI Study Assistant that:

1. Loads student performance data from a CSV file (Pandas)
2. Cleans and visualizes the data (Pandas + Matplotlib)
3. Trains a ML model to predict pass/fail (Scikit-Learn)
4. Stores course notes in a vector database (ChromaDB)
5. Answers questions about course content using RAG (AI API + ChromaDB)

---

2. Project Structure

ai-study-assistant/
├── data/
│   └── students.csv
├── notes/
│   └── python_notes.txt
├── .env
├── .gitignore
├── requirements.txt
├── step1_data_analysis.py
├── step2_ml_model.py
├── step3_rag_assistant.py
└── README.md

---

3. Step 1: Data Analysis

Create step1_data_analysis.py:

import pandas as pd
import matplotlib.pyplot as plt

# Load data
df = pd.read_csv("data/students.csv")

# Explore the data
print("Shape:", df.shape)
print("\nFirst 5 rows:")
print(df.head())
print("\nBasic stats:")
print(df.describe())
print("\nMissing values:")
print(df.isnull().sum())

# Clean: Fill missing values
df["study_hours"] = df["study_hours"].fillna(df["study_hours"].mean())
df["attendance"] = df["attendance"].fillna(df["attendance"].median())

# Visualize: Study hours vs Score
plt.scatter(df["study_hours"], df["score"], color="blue", alpha=0.6)
plt.title("Study Hours vs Exam Score")
plt.xlabel("Study Hours per Week")
plt.ylabel("Exam Score")
plt.savefig("scatter_plot.png")
plt.show()

# Visualize: Pass/Fail distribution
df["result"].value_counts().plot(kind="bar", color=["green", "red"])
plt.title("Pass vs Fail Distribution")
plt.savefig("pass_fail_chart.png")
plt.show()

print("Step 1 complete! Charts saved.")

---

4. Step 2: ML Model

Create step2_ml_model.py:

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score, classification_report

# Load and prepare data
df = pd.read_csv("data/students.csv")
df = df.dropna()

# Features and target
X = df[["study_hours", "attendance", "past_score"]].values
y = (df["score"] >= 50).astype(int).values  # 1 = Pass, 0 = Fail

# Split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Train
model = DecisionTreeClassifier(random_state=42)
model.fit(X_train, y_train)

# Evaluate
predictions = model.predict(X_test)
accuracy = accuracy_score(y_test, predictions)
print(f"Model Accuracy: {accuracy * 100:.1f}%")
print("\nDetailed Report:")
print(classification_report(y_test, predictions, target_names=["Fail", "Pass"]))

# Predict for a new student
new_student = np.array([[6, 80, 55]])  # 6 hours study, 80% attendance, 55 past score
result = model.predict(new_student)
print(f"\nNew Student Prediction: {'Pass' if result[0] == 1 else 'Fail'}")

---

5. Step 3: RAG Study Assistant

Create step3_rag_assistant.py:

import chromadb
import requests
import os
from dotenv import load_dotenv

load_dotenv()
api_key = os.getenv("GEMINI_API_KEY")

# Setup vector database with course notes
client = chromadb.Client()
collection = client.create_collection("study_notes")

# Add your course notes
notes = [
    "A list in Python stores multiple items in one variable. Use square brackets [].",
    "A dictionary stores data as key-value pairs using curly braces {}.",
    "A for loop repeats code a fixed number of times. Use range() for number sequences.",
    "Functions are reusable blocks of code. Define with def keyword.",
    "NumPy arrays are faster than lists for mathematical operations.",
    "Pandas DataFrames are like Excel tables for data analysis.",
    "Machine Learning lets computers learn patterns from data automatically.",
    "train_test_split divides data into training and testing sets.",
    "RAG stands for Retrieval-Augmented Generation. It adds context to AI prompts."
]

collection.add(
    documents=notes,
    ids=[f"note_{i}" for i in range(len(notes))]
)

def ask_assistant(question):
    # Step 1: Find relevant notes
    results = collection.query(query_texts=[question], n_results=3)
    context = "\n".join(results["documents"][0])

    # Step 2: Build RAG prompt
    prompt = f"""You are a friendly Python tutor. Answer the student's question using
the context provided. Keep your answer short and simple.

Context from course notes:
{context}

Student's question: {question}

Answer:"""

    # Step 3: Call Gemini API
    url = f"https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash:generateContent?key={api_key}"
    payload = {"contents": [{"parts": [{"text": prompt}]}]}
    response = requests.post(url, json=payload)
    answer = response.json()["candidates"][0]["content"]["parts"][0]["text"]

    return answer

# Interactive chat loop
print("AI Study Assistant (type 'quit' to exit)")
print("-" * 40)
while True:
    question = input("\nYour question: ")
    if question.lower() == "quit":
        print("Goodbye! Keep studying!")
        break
    answer = ask_assistant(question)
    print(f"\nAssistant: {answer}")

---

6. Supporting Files

requirements.txt:

pandas
numpy
matplotlib
scikit-learn
chromadb
requests
python-dotenv

.gitignore:

.env
__pycache__/
*.pyc

.env:

GEMINI_API_KEY=your-api-key-here

---

7. What to Include in Your GitHub README

# AI Study Assistant

An AI-powered study assistant built with Python that:
- Analyzes student performance data
- Predicts pass/fail using Machine Learning
- Answers questions about Python using RAG

## Technologies Used
- Python, Pandas, Matplotlib, Scikit-Learn
- ChromaDB (Vector Database)
- Google Gemini API (LLM)
- RAG (Retrieval-Augmented Generation)

## How to Run
1. Clone this repo
2. Install dependencies: `pip install -r requirements.txt`
3. Add your API key to `.env`
4. Run each step: `python step1_data_analysis.py`

---

Summary

You just built a complete AI application that combines:

Skill	Where You Used It
Python Basics	Variables, loops, functions, file handling
Pandas	Loading CSV, cleaning data, analysis
Matplotlib	Visualizing student data
Scikit-Learn	Training a classification model
APIs	Calling Google Gemini
RAG + ChromaDB	Building a smart Q&A assistant
Pydantic / dotenv	Secure API key handling
Git	Version control and README

This one project proves you can do the job of an AI/ML Engineer. Put it on your GitHub and add the link to your resume!