Predicting Credit Card Fraud with R

Offert par
Université du Nord Texas
Dans ce projet guidé, vous :

Use R to identify fraudulent credit card transactions with a variety of classification methods.

Create, train, and evaluate decision tree, naïve Bayes, and Linear discriminant analysis classification models using R

Generate synthetic samples to improve the performance of your models.

Clock1.5 hours
IntermediateIntermédiaire
CloudAucun téléchargement requis
VideoVidéo en écran partagé
Comment DotsAnglais
LaptopOrdinateur de bureau uniquement

Welcome to Predicting Credit Card Fraud with R. In this project-based course, you will learn how to use R to identify fraudulent credit card transactions with a variety of classification methods and use R to generate synthetic samples to address the common problem of classification bias for highly imbalanced datasets—the class of interest (fraud) represents less than 1% of the observations. Class imbalance can make it difficult to detect the effect independent variables have on fraud, ultimately leading to higher misclassification rates. Fixing the imbalance allows the minority class (fraud) to be better learned by the classifier algorithms. After completing the project, you will be able to apply the methods introduced in the project to a wide range of classification problems that typically confront class imbalance, including predicting loan default, customer churn, cancer diagnosis, early high school dropout risk, and malware detection. Note: This course works best for learners who are based in the North America region. We’re currently working on providing the same experience in other regions.

Les compétences que vous développerez

Data AnalysisMachine LearningR Programming

Apprendrez étape par étape

Votre enseignant(e) vous guidera étape par étape, grâce à une vidéo en écran partagé sur votre espace de travail :

  1. Task 1: Explore why imbalanced datasets are problematic for classification algorithms.

  2. Task 2: Use R to explore a dataset.

  3. Task 3: Create random testing and training datasets using the caret package in R.

  4. Task 4: Use R to synthetically balance your training dataset using three techniques from the smotefamily package.

  5. Task 5: Train three classification algorithms (decision tree, naïve Bayes, and linear discriminant analysis) using the natively imbalanced dataset, and generate the predictions for the test dataset.

  6. Task 6: Use R to visually compare your models using the recall, precision, and F measure classification accuracy metrics.

Comment fonctionnent les projets guidés

Votre espace de travail est un bureau cloud situé dans votre navigateur, aucun téléchargement n'est requis.

Votre enseignant(e) vous guide étape par étape dans une vidéo en écran partagé

Foire Aux Questions

Foire Aux Questions

D'autres questions ? Visitez le Centre d'Aide pour les Etudiants.