I have a historical data of orders from different customers of my company and delivery details. I wanted to develop a model which predicts the probability of customers order getting delivered before or on the "Requested Delivery Date".

In my data-set, i've details about customer delivery location product delivery type product classification (Can be exported or not) etc.

I was not able to start this model as I am not sure how to tackle this problem. It would be of great help, if you could suggest some reading material on the same or few link directing the same.

  • 2
    $\begingroup$ I have two options in mind: Use a regression model to predict number of days for product to arrive. Or use a classification model to predict whether or not a delivery while arrive before a given time. Look those terms up and start learning :) $\endgroup$
    – yoav_aaa
    Feb 1, 2018 at 9:44

2 Answers 2


The most important thing here is to model your problem as time series, as you know, each example consists of one customer that has many orders over time. see Predicting Clinical Events via Recurrent Neural Networks


It sounds like your dataset is going to need some work before you can start applying a model to it, for example you will have to translate customer location into something more usable, like distance from depot, or Urban/Suburb/Country labels perhaps?

Andrew Ng's Coursera course is a popular starting point, many beginners seem to find it useful. You'll get a good understanding of machine learning algorithms and how to arrange your data to get the best out of the models.

Good luck!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.