Back

Explore Courses Blog Tutorials Interview Questions
0 votes
2 views
in Data Science by (18.4k points)
I have applied for a data scientist post in a private company. The company has given me a case study, where I have to predict future expenses of the customers using the dataset. The dataset set consists of all the expenditures details of an individual customer. (Example:  time, receipt, amount).

As per my knowledge, I need to predict the monthly payments such as insurance, electricity bill and so on. But I have no idea which algorithm to be used. Can anyone suggest me which algorithm to use?

1 Answer

0 votes
by (36.8k points)

You are actually thinking in a correct way but the approach is ambiguous. and complex because the discount factor is a random variable.

Regular payments are easy to predict since they have a strict schedule, for example, EMI. and random payments which can be distributed differently. What I meant to say is you need to build 2 different models. One for the 'regular' payment and another model for 'random' payment.

Using Poisson distribution we can build the irregular payment model. The parameters hear can be inferred from the history and their distribution has high variance. We need to assume that their distribution doesn't vary.

For regular parameters, we can use the Time series algorithm. Try finding the patterns in the dataset. If you find patterns then use the seasonal ARIMA model.

So you will be having two models in hand one for regular payment and another for irregular payments. Using them you can predict the expenses of an individual customer.

You can use the link Data Science for more information on the subject and improve your knowledge

Browse Categories

...