Data quality

There’s a variety of data quality checks that can be implemented as business rules. With our solution, data analysts and engineers can create rules to ensure that certain data columns don’t exceed pre-defined ratios of nulls, validate that data falls into certain ranges, or check that a data set complies with a certain profile. The tool assists with data profiling, measuring data quality metrics, cleansing and auto-correcting data, and alerting the support team when something goes wrong.

If your data analytics platform already has thousands of data processing jobs or the business rules being used aren’t detecting complex data defects, anomaly detection can help build a more comprehensive data quality solution. Data scientists can configure automatic data profiling to collect key data metrics, use statistical process control techniques, and configure deep learning anomaly detection to uncover suspicious patterns and alert the support team if predefined levels of confidence are reached.

Good data quality starts with ensuring that the raw data imported into the data analytics platform is done correctly and completely, is consistent, and not stale. With our solution, we can configure various types of checks that integrate with data sources in data lakes or SQL-based databases. Measuring and improving data completeness is critical for streaming use cases such as clickstream processing, order processing, payment processing, or Internet of Things applications, when events can be dropped or processed more than once.