Build a predictive model

This is phase one of a project to determine the expected payment date for a very large pool of delinquent receivables which are being acquired. The estimated payment period is based on the debtors’ historical payment pattern and will be determined for each individual debtor.

Each debtor must pay the receivable during a 24-month period and we have a 10-year payment history. The payment history indicates for each year of origination for the number of months the receivable, whether the receivable became delinquent and was sold, and if the receivable remains unpaid. An example of the available data is below.

Year Example #1 Example #2 Example #3 Example #4 Example #5 Example #6
2022 12 U 6 5 12 U
2021 24 12 18 4 12 23
2020 24 12 18 5 15 20
2019 24 10 24 5 15 23
2018 24 5 N 4 16 21
2017 24 6 N 4 18 22
2016 24 6 N 10 15 24
2015 24 5 6 8 24 23
2014 24 6 6 1 24 20
2013 24 10 5 2 23 21

U = Receivable unpaid at determination date
N = Receivable paid during regular payment period and not sold

Step 1 of this phase will be to determine what number of historical payments is most indicative of future payments.

Step 2 of this phase will be to develop a predictive model to forecast the future payments based on the period determined in Step 1.

We are looking for a person with the ability to apply statistical techniques to analyze data and identify patterns and trends; the ability to apply machine learning algorithms to develop a predictive model; and proficiency in programming languages such as Python, R or Matlab.

This is NOT a writing project. WRITERS PLEASE DO NOT APPLY.