Titanic_Survival_Predictor

Problem Statement:

Predict survival on the Titanic using logistic regression by exploring relationships between passenger characteristics and survival outcomes through data cleaning, EDA, and model training. Dataset sourced from Kaggle.

Objective:

To predict Titanic survival with a logistic regression model, aiming for higher accuracy, precision, and recall using dataset features.

Libraries Used:

1. Numpy (for data manipulation)
2. Pandas (for data manipulation)
3. Matplotlib (for data visualization)
4. Seaborn (for data visualization)
5. Statstools (for data modeling)
6. Scikit-Learn (for data modeling)
7. Collections (for counting occurences)
8. imblearn (for oversampling)

Contents:

1. Importing Libraries
2. Importing Dataset
3. Data Understanding
4. Handling Missing Values

• 4.1 Handling missing values - Dropping
• 4.2 Handling missing values - Imputing
  • 4.2.1 Feature engineering

5. Exploratory Data Analysis

• 5.1 Bivariate Analysis
• 5.2 Multivariate Analysis

6. Data Preparation for Modeling

• 6.1 Binary Encoding
• 6.2 Splitting train and test data set
• 6.3 Resampling Class Imbalance
• 6.4 Feature Scaling

7. Training the Model

• 7.1 Model Creation
• 7.2 VIF

8. Precision - Recall Analysis

• 8.1 Confusion Matrix
• 8.2 Precison and Recall
• 8.3 Optimal Cut off - Precision_Recall_Curve

9. Predicting on the Test Data
10. Preparation and Submission

• 10.1 Missing values
  • 10.1.1 Handling missing values - Imputing
• 10.2 Binary Encoding
• 10.3 Scaling the Features
• 10.4 Dropping the unnecessary columns
• 10.5 Prediction and Submission