Predicting Farmer Income in India

Access to credit is the key to attaining full economic potential. People employed in farming and allied activities struggle to get loans due to insufficient or non-existent credit histories. As a result, this population is often exploited by untrustworthy lenders. L&T Finance (LTF) aims to broaden financial inclusion for the unbanked population by providing a positive and safe borrowing experience. To ensure a safe borrowing experience, LTF utilizes several data sources to assess the creditworthiness of the population. In addition to traditional credit data vendors, we leverage weather monitoring datasets, soil quality data, crop yield, commodity price dynamics, and land records to predict their client’s ability and willingness for repayment.

Although LTF uses state-of-the-art models that leverage statistical and machine learning models along with emergent GenAl literature to profile and predict the creditworthiness of the farming population. We challenge economic modelling community to help us unlock the full potential of our data assets and raise our methods of assessing creditworthiness to a level where the loan application of a farming professional worthy of repayment should not get rejected. We also encourage the teams to acquire and use alternate public or private data sources that may help improve the credit assessment model.

The solution will be evaluated based on MAPE (mean absolute percentage error) of predicted farmer income on unseen data.