Data vs Information: What’s The Difference?

Here’s why it matters more than you think

Since the dawn of internet people have been using data and information interchangeably. The world of data is becoming more and more relevant with the rise of AI, yet people still confuse these terms. The average person gets a pass, but data professionals shouldn’t.

Information is derived from knowledge. Knowledge is derived from activities (measurements, thoughts, debates, etc). This would imply that information is always being created. From neurons to sensors.

The problem is that capturing information is not always so simple (much less communicating). The whole point is being able to communicate knowledge for the benefit of others or for storage. Because if we already know something, what’s the point of keeping to ourselves? One way of doing it is encoding our information. And that is data.

Example

There’s nothing better than a good ol’ example. In this case let’s imagine the story of a marketing company. Every month this company runs a different marketing campaign through various media (online, tv, magazines, etc). These campaigns generate information, thus creating data.

Tables with data are called datasets. Datasets have rows and columns that give us an understanding of the table. Rows are instances of associated information. In this example, each row would be a marketing campaign. Columns represent specific information. In this example we can compare each marketing campaign on specific information like profit or location.

Rows are also called observations, records, tuples or trials. Columns are also called features, fields, attributes, predictors or variables.

As data professionals, we should familiarize ourselves with our workplace’s lingo. Different industries use different terms for the same concepts, and it’s important to know them so we can keep up in conversations.

Data points are intersections in tables. In this example, a data point could be that 100 units were sold on Halloween. It’s also important to mention that datasets have headers that indicate what each feature means, and paint the big picture. If headers are not included, it should be implied.