Читать книгу Blockchain Data Analytics For Dummies - Michael G. Solomon - Страница 15

One of the obstacles to realizing the full value of data is the dependence on its quality. Quality data is valuable, while incomplete or untrusted data is often worthless. What’s worse, low-quality data may require more budget to clean than it will potentially generate in revenue. The only way to realize data’s true value is to ensure that the data is valid and represents entities in the real world.

Verifying data has long been one of the highest costs associated with collecting and using data. Campaigns that depend on physical or email addresses will have little effect if the target addresses are largely incorrect. Bad data can come from many sources, including mischievous data submission, sloppy data collection, or even malicious data modification. An important aspect of relying on data is putting controls in place that verify the source of any collected data, along with that data’s adherence to collection requirements.

A simple approach to verifying data in a distributed environment is to carry out a simple validation at the source and again at the server as the data is stored in a repository. While validating data at least twice may seem excessive, the practice makes user errors easier to catch and ensures that data received by the server is clean.

Validating data twice makes it possible for client applications to quickly catch errors, such as too many digits in a phone number or a missing field, while the server handles more complex validation tasks. A server may need access to other related data to ensure that data is valid before storing it in a repository. Server validation could include things such as verifying that order quantities are available in a warehouse and that data wasn’t changed by a malicious agent during transmission from the client.

One of the reasons data verification is so important is that organizations are relying more and more on their data to direct business efforts. Aligning business activities with expectations based on faulty data leads to undesirable results. In other words, decisions are only as good as the data on which those decisions are based. The “garbage in, garbage out” adage still holds true.