Showing 1–1 of 1 results for author: Ots, H
-
Mobile Phone Usage Data for Credit Scoring
Authors:
Henri Ots,
Innar Liiv,
Diana Tur
Abstract:
The aim of this study is to demostrate that mobile phone usage data can be used to make predictions and find the best classification method for credit scoring even if the dataset is small (2,503 customers). We use different classification algorithms to split customers into paying and non-paying ones using mobile data, and then compare the predicted results with actual results. There are several re…
▽ More
The aim of this study is to demostrate that mobile phone usage data can be used to make predictions and find the best classification method for credit scoring even if the dataset is small (2,503 customers). We use different classification algorithms to split customers into paying and non-paying ones using mobile data, and then compare the predicted results with actual results. There are several related works publicly accessible in which mobile data has been used for credit scoring, but they are all based on a large dataset. Small companies are unable to use datasets as large as those used by these related papers, therefore these studies are of little use for them. In this paper we try to argue that there is value in mobile phone usage data for credit scoring even if the dataset is small. We found that with a dataset that consists of mobile data based only on 2,503 customers, we can predict credit risk. The best classification method gave us the result 0.62 AUC (area under the curve).
△ Less
Submitted 28 February, 2020;
originally announced February 2020.
