W1_Quiz_Q7_Why audio can't be bracketed as structured data?

“A dataset is composed of age and weight data for several people. This dataset is an example of “structured” data because it is represented as an array in a computer.”

an audio record is also an array in a computer so why it’s no a reason for be structured?

Hello Gregory,

Welcome to the community.

An audio file or a data is by nature an array. They are categorized as unstructured data as they get changed with the change in technology. They do not comply as a well-proven relational structures with time. I think this is one of the main reasons why an audio record being an array can’t be bracketed under structured data.

In my mind the distinction is that for 2D structured data, the dimensions mean profoundly different things. In the person/age/weight example, the horizontal axis is a set of characteristics or attributes: age, weight, name, date-of-birth etc. Whereas the vertical represents distinct entities, each of which is represented with those same attributes. You could think of it as columns carry the state for the object represented by that row. The rows are records, the columns are the schema, or structure, of each record.

Now consider a 2D image. The array of pixels doesn’t have a different meaning between different rows or columns. Moving sideways one pixel doesn’t change it to the ‘age’ pixel, or the ‘date-of-birth’ pixel. Moving up or down doesn’t change the context - the pixels there cannot in any real sense be considered to belong to or comprise the state of that row. Since all the pixels in both dimensions are equal in meaning, there is no schema, or structure for them. Notice if a picture was associated with each record in the person database, the image wouldn’t be stored as rows and columns of pixel values, but as a blob.

Based on that idea, is audio more like the person database or image? Possibly it depends on the exact audio file format, since you could argue that something like an MP3 header has structure. But in questions like this, people are most often referring to the audio data itself, the unstructured stream of numeric values.

Does that make sense?

PS: in my opinion a person-age-weight dataset is structured regardless of how it is represented in the computer. Array, graph network, compound XML object. Doesn’t matter. What matters is that the three values have specific meaning and are related in a specific way, namely person has-a age, and person has-a weight.Can’t say the same for pixels in an image. Cheers. @paulinpaloalto I added this PS after your endorsement so fair warning if you need to retract :grin:

Hi ai_curious,

While I agree with your explanation, I think that this question should be rephrased. As it is, I would answer False, because although the age-weight dataset is structured, that is not a consequence of it being representable as an array (as the question implies), since an audio clip could also be represented as such, while not being structured.

I would simply remove the “because” and everything else up to the full stop from the question statement.

I took this course in 2017 so have long ago lost the ability to see the quiz questions. And never had any authority on how they were expressed or evaluated. Was merely trying to add my perspective to whether audio data files should be considered structured or unstructured data. Suggestions about the course content should be attention of DLAI staff. Cheers