Titanic_Prediction_Project deployed on streamlit 🔗 Try the App Now!

CodeSoft Internship Project 1

Project Overview

This was a Mchine learning project to predict whether a passenger on the Titanic survived or not using the famous Titanic dataset from Kaggle

Data Set

Source	Titanic Dataset by YasserH

Description

• Source: Titanic dataset (publicly available)
• Rows: 891 passenger records
• Columns: 12 columns, including:
  • PassengerId: Unique ID for each passenger
  • Survived: Target variable (0 = Did not survive, 1 = Survived)
  • Pclass: Ticket class (1 = 1st, 2 = 2nd, 3 = 3rd)
  • Name, Sex, Age, SibSp, Parch, Ticket, Fare, Cabin, Embarked

---

Exploratory data analysis

Data cleaning and preprocessing included:

• Dropping irrelevant columns like Name, Ticket, and Cabin
• Filling missing Age values with median age
• Filling missing Embarked values with mode
• Encoding categorical variables (Sex, Embarked) via one-hot encoding

---

Development Workflow

Step 1: Data Exploration & Model Building in Google Colab

• Load dataset CSV into pandas
• Perform exploratory data analysis to understand data quality and distributions
• Clean data by handling missing values
• Convert categorical variables to numeric via one-hot encoding
• Train a Logistic Regression model using scikit-learn with train-test split
• Achieved an accuracy of approximately 80.4%
• Save the trained model locally as titanic_model.pkl
• Download the model file from Colab to local machine

Step 2: Local Development - Virtual Environment Setup

• Create a project folder locally (e.g., Titanic_Prediction_Project)

• Move the downloaded titanic_model.pkl file into this folder

• Open Command Prompt (Windows Terminal) and navigate to the project folder

• Create and activate a Python virtual environment:

• Upgrade pip and install necessary libraries:

Step 3: Building the Streamlit App in VS Code

• Use VS Code to open the project folder (E:\Titanic_Prediction_Project)
• Create app.py in the folder
• Develop the Streamlit app that:
  • 🔄 Loads the saved model (titanic_model.pkl) using joblib
  • 🎛️ Provides input widgets for passenger features:

    Widget	Feature
    number_input	Passenger ID
    selectbox	Pclass (1,2,3), Sex, Embarked
    slider	Age, SibSp, Parch, Fare

  • 🧹 Processes input features with same preprocessing as training:
    • Fill missing Age with median
    • One-hot encode Sex, Embarked (drop_first=True)
    • Match exact column order: ['PassengerId', 'Pclass', 'Age', ...]
  • 🎯 Predicts survival with model, showing:
    • ✅ Result: "Survived" or "Did Not Survive"
    • 📊 Probabilities: Survival % + Death %

Code handles edge cases automatically - exact feature matching with training data!

Step 4: Running the App Locally

• Run the Streamlit app in terminal:
  • I was able to interact with the UI and get survival prediction in real time without needing the dataset

Step 5: Deploy on Streamlit

• Created repo on github
• Deplyed project on streamlit using the repo

---

Project Outcome

• Completed end-to-end ML project from data cleaning to deployment
• Gained practical experience working with real-world dataset and production tools
• Developed strong skills in data preprocessing, model training, and Python-based web app development
• Demonstrated ability to deliver a working, interactive application for real users

---

✨ Live Demo

🔗 Try the App Now!

80.4% Accuracy Model | Interactive Predictions | Deployed on Streamlit Cloud

---

How to Run This Project

1. Clone this repository.
2. Create and activate virtual environment.
3. Install dependencies: pip install -r requirements.txt
4. Run the app