Machine-Learning-Credit-Scoring-Geospatial

Research work on credit scoring using geospatial data and techniques

This practicum presents interactive dashboard for enhancing credit analysis and assessment using GeoSpatial techniques for Irish residential properties developed by data from following sources (i) loan portfolios and credit data (ii) property price register (iii) Central Statistical Office. These analyses can be used to reduce the chances of financial loss on a residential mortgage. The predictive model is built on logistic regression and decision tree algorithms and produces an estimated default probability of the applicant. Models are built on normalised data to cover all possible scenarios from real life. The predictive probability will determine good and bad customers by classifying them into four categories. The output from predictive models is used on tableau to generate business dashboard. Models ability to classify and performance mea- surements were measured by using statistical metrics: Gini, KS and AUROC.

Results showed that decision tree has much better performance than logistic regression.

Keywords: Credit Scoring, Logistic Regression, Decision Tree, Tableau, Auditing

a relative link

Final Dashboard <iframe src=”https://public.tableau.com/shared/3P3GFNMNJ?:showVizHome=no&:embed=true width=”90%” height=”500”></iframe> [iframe src=”https://public.tableau.com/shared/3P3GFNMNJ?:showVizHome=no&:embed=true width=”90%” height=”500”]

            <script type='text/javascript'>                    var divElement = document.getElementById('viz1520195417583');                    var vizElement = divElement.getElementsByTagName('object')[0];                    vizElement.style.width='1024px';vizElement.style.height='877px';                    var scriptElement = document.createElement('script');                    scriptElement.src = 'https://public.tableau.com/javascripts/api/viz_v1.js';                    vizElement.parentNode.insertBefore(scriptElement, vizElement);                </script>