Objective: In this assignment, you will create a logistic regression model to classify transactions as either belonging to a "new" cardholder or an "existing" cardholder based on engineered spending behavior features. You$$ll develop and evaluate your model using several key steps to understand its effectiveness and identify improvements.

Dataset: Credit card fraud detection dataset: cc$\_$info.csv$,$ transactions.csv Instructions:

$1.$ Feature Engineering $(10$ Points$)$: Engineer at least three new features that may help differentiate new cardholders from existing ones.

$$ Be creative with your features, but also justify why you believe they would help in classification.

$$ Add comments explaining each feature, why it is relevant, and how it is calculated.

$2.$ Model Training $(5$ points$)$

$$ Define the target variable new$\_$cardholder $(1$ for new, $0$ for existing$).$

$$ Split the data into training and test sets $(70\%$ training, $30\%$ test$).$

$$ Train a logistic regression model on the training set using the engineered features.

$$ Display the model summary, including coefficients and p$-$values, and explain what these mean.

$3.$ Model Evaluation $(5$ points$)$

$$ Predict on the test set and generate the following evaluation metrics:

$$ Confusion Matrix

$$ Accuracy, Precision, Recall, and F$1$ Score $$ Explain what each of these metrics means in the context of the problem.

$4.$ Model Diagnosis and Improvement $(5$ points$)$ If the model does not perform well $($i$.$e$.,$ low accuracy, over$/$under fitting or imbalanced precision$/$recall$)$:

$$ Identify possible reasons for poor performance $($e$.$g$.,$ class imbalance, weak features$).$

$$ Apply methods to improve the model $($e$.$g$.,$ oversampling, undersampling, or feature selection$).$

$5.$ Conclusion $(5$ points$)-$ Summarize your approach, findings, and any suggestions for further improvement.