class: center, middle, inverse, title-slide .title[ #
Analyzation of Credit Approval rates using logistic regression.
] .subtitle[ ##
] .author[ ###
Kyle Weber
] .institute[ ###
West Chester University of Pennsylvania
] .date[ ###
Prepared for
STA490: Data Visualization
Slides available at:
https://rpubs.com/KW986324
AND
https://github.com/Kyle-Weber/STA490
] --- class: top, center # Table of Contents .left[ 1. Title Slide <br> <br> 2. Data Set Summary <br> <br> 3. Research Question <br> <br> 4. Exploratory Analysis <br> <br> 5. Full Logistic Regression Model <br> <br> 6. Reduced Logistic Regression Model <br> <br> 7. Final Logistic Regression Model <br> <br> 8. Model Comparison <br> <br> 9. Odds Ratio Analysis <br> <br> 10. Conclusion ] --- class: top, center # Research Question <br> .left[ <font size="5"> - Can a multiple logistic regression model predict credit card approval based on applicant’s demographic and financial information? </font> ] --- class: top, center # Exploratory Analysis .left[ - Pairwise Scatter Plots for Numerical Variables <br> - There are issues regrading skewness and grouping <br> - grouping the histograms will help to fix this problem ] <img src="SoloP_files/figure-html/interactive-scatterplot-matrix-1.png" width="100%" /> --- class: top, center # Full Logistic Regression Model .left[ - Full Model <br> <br> - Helps to find variables that are not statistically significant ] <div class="figure">
<p class="caption">Summary of inferential statistics of the full model</p> </div> --- class: top, center # Reduced Logistic Regression Model .left[ - Reduced Model <br> <br> - non-significant variables taken out manually ] <div class="figure">
<p class="caption">Summary of inferential statistics of the reduced model</p> </div> --- class: top, center # Final Logistic Regression Model .left[ - Final Model <br> - automatic variable selection <br> - most relevant variables kept in ] <div class="figure">
<p class="caption">Summary of inferential statistics of the final model</p> </div> --- class: top, center # Goodness-of-fit Model Comparison .left[ - Comparison of three logistic regression models. <br> <br> - final model has a slightly better Deviance and AIC ] <br> <br> <div class="figure">
<p class="caption">Comparison of global goodness-of-fit statistics</p> </div> --- class: top, center # Odds Ratio analysis .left[ - performed for a more practical interpretation <br> - Income group "0.1-5" has odds ratio 0.1388257, indicating about 86% lower approval odds than the baseline <br> - Similar patterns observed for other variables <br> - Priordefault is the most influential variable ] <div class="figure">
<p class="caption">Summary Stats with Odds Ratios</p> </div> --- class: top, center # Conclusion .left[ - Multiple logistic regression serves as a powerful tool for extracting insights from intricate datasets. - Offers a deeper understanding of the dynamics influencing credit card approval decisions. - Revealed crucial information about the relationship between credit card approval and predictor variables. ]