Building a churn predictor using Hadoop

In this section, we will apply the knowledge gained in the previous section to build a churn predictor that can scale up for use with very large datasets. Hadoop is particularly suited to the processing of very large datasets.

Synthetic data generation tools

To build our churn prediction using Hadoop, we will work with a synthetic dataset. Synthetic datasets are good for testing and learning purposes because they do not carry the risk of personal information leakage, and they are simple to obtain or create. But they also carry the limitation that, unlike a real-life dataset, they never contain the truth. Synthetic data offers a good alternative to real life datasets if you want to get started quickly.

Some synthetic data generation techniques and tools are covered as follows:

Note

Synthetic Data Generation

To generate synthetic datasets, you can use a site such as http://www.generatedata.com/, which offers the generation of 100 rows in a dataset for free...