Summary of Data Science Internship Projects at ArchTechnologies

1. Task 1: Titanic Survival Prediction (Classification)

Problem Type: Binary Classification
Model Used: Logistic Regression
Goal: Predict whether a passenger survived based on features like Gender, Age, Passenger Class (PClass), and Fare.

Key Result: Achieved approximately 78.3% accuracy.
Key Findings:

• Gender was the most significant predictor (females had higher survival rates).
• Passenger Class strongly impacted likelihood of survival (1st-class passengers survived more often).

---

2. Task 2: Stock Price Prediction (Time Series Forecasting)

Problem Type: Time Series Regression
Model Used: LSTM Neural Network (Long Short-Term Memory)
Goal: Predict the next day's closing stock price using sequences of historical OHLCV data (Open, High, Low, Close, Volume).

Key Result: Successfully predicted next-day closing price.
Performance Evaluation:

• Mean Squared Error (MSE)
• R² Score

---

3. Task 3: Customer Segmentation (Unsupervised Learning)

Problem Type: Unsupervised Clustering
Model Used: K-Means Clustering
Goal: Segment retail customers using Age, Annual Income, and Spending Score.

Methodology:

• Optimal number of clusters determined using the Elbow Method → k = 5

Key Result: Identification of 5 distinct customer segments, such as:

• High-income Luxury Spenders
• Budget-conscious Shoppers
• High-income Careful Spenders

---

4. Task 4: IMDB Movie Rating Predictor (Regression)

Problem Type: Regression
Model Used: Random Forest Regressor
Goal: Predict final IMDB Rating using Votes, Metascore, Revenue (Millions), Runtime, and Genre.

Key Results:

• R² = 0.7410
• RMSE = 0.4312

Feature Importance:

• Votes and Metascore were the most influential features.

---

Technologies Used

Core Languages

• Python

Data Manipulation & Analysis

• pandas
• NumPy

Visualization

• Matplotlib
• Seaborn

Machine Learning (Traditional)

• scikit-learn

Deep Learning (Time Series)

• TensorFlow / Keras (LSTM)