The world’s leading publication for data science, AI, and ML professionals.

How to ace cold calling with Machine Learning

To Call or not to Call?

Cold Caller: "Hello Luke, how are you doing today? I noticed you were up for car insurance renewal and I was wondering whether you would have a couple of minutes to hear about this amazing offer I can give you to reward you for your loyalty?"

What would you do if you were Luke?

It’s actually very hard to tell, which is why cold calls have an extremely low success rate and require thousands of calls to bring any value. Because of the randomness involved they are not really pleasant for either parties involved. However, with Machine Learning and enough data it might be possible to understand the factors behind successful calls which can then be used to better qualify prospects and tailor the content of the calls.

In this article we are going to build a machine learning model for predicting the success of car insurance cold calls using AuDaS, the Automated Data Science team in a box. The data was provided by the Technical University of Munich and is hosted on Kaggle.

Data Exploration

The dataset was collected from a bank in the United States that also offers car insurance and contains information on their customers. The goal is therefore to build a model that can predict whether a customer will subscribe for Car Insurance or not. A quick overview of this information is provided in the following table:

To understand the features driving conversions we can try to visualise how the duration of the cold calls impacts the outcome. First we need to create the feature through AuDaS’ Data preparation page by adding a datediff operation.

Data Preparation in AuDaS
Data Preparation in AuDaS

We can then access the histogram view to eye ball the possible relationships.

Histogram view in AuDaS
Histogram view in AuDaS

Surprisingly, although the call duration does increase the chance of conversion (CarInsurance = 1), there doesn’t seem to be a clear relationship. The hour of the call doesn’t seem to affect the outcome either.

Distribution of successful conversions across all the features
Distribution of successful conversions across all the features

Overall there do seem to be patterns in our dataset which is why we can be hopeful that AuDaS will find a Machine Learning model that will perform well!

Automated Modelling

Our goal is to build a classification pipeline to predict whether a customer will purchase Car Insurance or not. However, we need to be able to use this model in production which is why we are going to exclude the Duration of the call and the End time stamp of the call as this is information we can not anticipate before the call!

For the model training, AuDaS will perform a 10-fold cross validation and will withhold a balanced 10% sample of the original data for final validation purposes. AuDaS will also use OPTaaS, Mind Foundry‘s proprietary Bayesian Optimiser to efficiently navigate the space of possible data science pipelines.

AuDaS provides the full transparency of the best found Data Science pipeline (models, parameter values,…) as well as score metrics.

Score metrics of the best found model
Score metrics of the best found model

Once the runs are complete, we can view the performance of the model on the 10% hold out and we are reassured to see that the classification accuracy is pretty good (72.4%) and that the model health is good!

AuDaS Model Health Diagnostic
AuDaS Model Health Diagnostic

Interpreting the model

The feature relevance of our model indicates that the yearly average balance of the customer’s bank account seems to have the strongest impact on them purchasing the car insurance or not, followed by their age and the time of the cold call.

AuDaS feature relevance
AuDaS feature relevance

In other words, if Luke was 40, with a high balance on his account there would be a strong chance of him taking the cold callers offer!

This model can then be put into production automatically by AuDaS through a RESTful API.

This means that the cold callers could use the machine learning model trained by AuDaS to prioritise the customers they should call which will help them increase the success of the marketing campaign. Unfortunately, this data set doesn’t provide enough insights to tailor the content of these calls but we could imagine that analysing the transcripts might allows us to identify some interesting recommendations!

A full video of this task with AuDaS can be viewed bellow. Please don’t hesitate to reach out by email or LinkedIn should you wish to try AuDaS.

AuDaS

AuDaS is an Automated Data Science platform developed by Mind Foundry that provides a robust framework for building end-to-end Machine Learning solutions (Classification, Regression, Clustering and soon Time Series) . This framework helps identify and act on data leakage before it’s too late. You can read some other practical use cases [here](https://towardsdatascience.com/solving-the-kaggle-telco-customer-churn-challenge-in-minutes-with-audas-2273fed19961) and here.

Sign up to try AuDaS:

[UPDATE: I have started a tech company. You can find out more here]

Team and Resources

Mind Foundry is an Oxford University spin-out founded by Professors Stephen Roberts and Michael Osborne who have 35 person years in data analytics. The Mind Foundry team is composed of over 30 world class Machine Learning researchers and elite software engineers, many former post-docs from the University of Oxford. Moreover, Mind Foundry has a privileged access to over 30 Oxford University Machine Learning PhDs through its spin-out status. Mind Foundry is a portfolio company of the University of Oxford and its investors include Oxford Sciences Innovation, the Oxford Technology and Innovations Fund, the University of Oxford Innovation Fund and Parkwalk Advisors.


Related Articles