American Express Credit Risk Analyst Interview Questions and Answers

Assess credit worthiness by analyzing credit history, financial statements, and other relevant data.

• Review credit reports and scores

• Analyze financial statements and income

• Consider employment history and stability

• Evaluate debt-to-income ratio

• Assess collateral and assets

• Look for red flags such as bankruptcies or late payments

Outlier analysis involves identifying and treating data points that are significantly different from the rest.

• Identify outliers using statistical methods such as box plots, scatter plots, and z-scores.

• Determine the cause of the outlier and decide whether to remove it or keep it in the dataset.

• Consider the impact of outliers on the analysis and adjust the model accordingly.

• Use techniques such as winsorization or data tr...read more

Multiple imputation can be used to impute missing values by creating multiple datasets with imputed values.

• Use multiple imputation to create multiple datasets with imputed values

• Combine the results from the multiple datasets to obtain a final imputed dataset

• Consider using predictive models to impute missing values

• Evaluate the quality of imputation using metrics such as mean squared error or R-squared

Variable selection can be done using techniques like correlation matrix, stepwise regression, and principal component analysis.

• Check for correlation between variables using correlation matrix

• Use stepwise regression to select variables based on their significance

• Perform principal component analysis to identify important variables

• Check for multicollinearity using variance inflation factor (VIF)

• Consider domain knowledge a...read more

Variable importance can be tested using various methods such as permutation importance, drop column importance, and SHAP values.

• Permutation importance involves randomly shuffling the values of a variable and measuring the decrease in model performance.

• Drop column importance involves removing a variable from the model and measuring the decrease in model performance.

• SHAP values provide a measure of the contribution of ea...read more

Logistic Regression is a statistical method used to analyze and model the relationship between a binary dependent variable and one or more independent variables.

• It is used when the dependent variable is binary (0 or 1).

• It estimates the probability of an event occurring based on the values of the independent variables.

• It is commonly used in credit risk analysis to predict the likelihood of default.

• It can also be used in...read more

Model performance of classification models can be tested using various metrics.

• Use confusion matrix to calculate accuracy, precision, recall, and F1 score.

• ROC curve and AUC can be used to evaluate model's ability to distinguish between positive and negative classes.

• Cross-validation can be used to test model's performance on different subsets of data.

• Use lift charts to compare model's performance with random selection.

• U...read more

Loss function in Logistic Regression measures the difference between predicted and actual values.

• It is used to optimize the model parameters during training.

• The most common loss function used in logistic regression is the binary cross-entropy loss.

• The goal is to minimize the loss function to improve the accuracy of the model.

• The loss function is calculated using the predicted probabilities and the actual labels.

• Other l...read more

Xgboost is a gradient boosting algorithm used for classification and regression tasks. It is faster and more accurate than Random Forest.

• Xgboost stands for Extreme Gradient Boosting

• It is a type of gradient boosting algorithm that uses decision trees

• It is faster and more accurate than Random Forest

• Xgboost uses a more regularized model formalization to control overfitting

• Random Forest builds multiple decision trees and c...read more

The loss function used in Xgboost is customizable and can be specified by the user.

• Xgboost supports various loss functions such as binary logistic regression, multi-class classification, and regression.

• The default loss function for binary classification is logistic regression while for regression it is mean squared error.

• Users can specify their own loss function by defining a custom objective and evaluation function.

• Th...read more

Xgboost parameters include learning rate, max depth, subsample, colsample by tree, and more.

• Learning rate controls the step size during training.

• Max depth limits the depth of each tree.

• Subsample controls the fraction of observations to be randomly sampled for each tree.

• Colsample by tree controls the fraction of features to be randomly sampled for each tree.

• Other parameters include min child weight, gamma, and lambda fo

Learning rate controls the step size at each boosting iteration in Xgboost.

• Learning rate is a hyperparameter that determines the contribution of each tree in the final output.

• A smaller learning rate requires more trees to be added to the model, but can lead to better performance.

• A larger learning rate can speed up the training process, but may result in overfitting.

• Typical values for learning rate range from 0.01 to 0....read more

The relationship between two features can be measured using correlation coefficient.

• Calculate the correlation coefficient using statistical methods.

• Correlation coefficient ranges from -1 to 1.

• A positive correlation indicates a direct relationship between the features.

• A negative correlation indicates an inverse relationship between the features.

• A correlation coefficient of 0 indicates no relationship between the feature

P value is the probability of obtaining a result as extreme or more extreme than the observed result, assuming the null hypothesis is true.

• P value is used in hypothesis testing to determine the significance of a result.

• A small p value (less than 0.05) indicates strong evidence against the null hypothesis.

• A large p value (greater than 0.05) indicates weak evidence against the null hypothesis.

• P value should not be used a...read more

Program to check if 4 coordinates form a square

• Calculate distance between all pairs of points

• Check if all distances are equal

• Check if diagonals are equal

• Use Pythagorean theorem to calculate distance