Encryptix Internship Projects

This repository contains the three projects completed as part of the internship program at Encryptix.

Project 1: Titanic Survival Prediction

Objective

Predict the survival of passengers on the Titanic using various machine learning techniques.

Steps Involved

1. Data Preprocessing:

   • Handled missing values
   • Feature engineering

2. Exploratory Data Analysis (EDA):

   • Analyzed feature distribution and correlations
   • Visualizations using bar charts, histograms, and heatmaps

3. Model Building:

   • Tested models: Logistic Regression, Decision Trees, Random Forest, Support Vector Machines
   • Used cross-validation for robustness

4. Model Evaluation:

   • Metrics: accuracy, precision, recall, F1 score
   • Hyperparameter tuning using Grid Search and Random Search

5. Result Interpretation:

   • Final model predictions and feature importance analysis

Project 2: Sales Prediction Using Python

Objective

Predict future sales of a company using historical sales data.

Steps Involved

1. Data Collection and Preprocessing:

   • Gathered and cleaned historical sales data
   • Encoding categorical variables

2. Time Series Analysis:

   • Analyzed trends, seasonality, and cyclic patterns

3. Feature Engineering:

   • Created features like moving averages, lagged features, rolling statistics

4. Model Building:

   • Explored ARIMA, Prophet, Random Forest, Gradient Boosting models

5. Model Evaluation:

   • Metrics: MAE, MSE, RMSE
   • Fine-tuning the best model

6. Deployment:

   • Developed a pipeline for continuous prediction updates
   • Visualizations and dashboards for stakeholders

Project 3: Credit Card Fraud Detection

Objective

Detect fraudulent credit card transactions to prevent financial losses and protect customers.

Steps Involved

1. Data Collection and Preprocessing:

   • Used a dataset of credit card transactions
   • Addressed data imbalance with SMOTE and other techniques

2. Handling Imbalanced Data:

   • Applied oversampling and undersampling

3. Feature Engineering:

   • Created features to capture transaction patterns
   • Scaling and normalization

4. Model Building:

   • Tested models: Logistic Regression, Decision Trees, Random Forest, Gradient Boosting, Neural Networks

5. Model Evaluation:

   • Metrics: Precision, Recall, F1 Score, AUC-ROC
   • Focused on minimizing false negatives

6. Implementation:

   • Real-time transaction monitoring and flagging suspicious activities
   • System for ongoing model updates and retraining