BSAN2201 Chap.2 Data Types and Management
Data Types and Management
Data classification determines which summaries and models are defensible. Numeric data divides into discrete counts and continuous measurements, while non-numeric data divides into nominal labels and ordinal rankings. Structured and unstructured data require different processing pipelines.
This chapter then follows the big-data management process from generation and collection through cleaning, structuring, analysis and communication. Data quality, privacy and security are treated as parts of analytical design, because evidence that is inaccurate, biased or improperly used cannot support a durable decision.
The same characters can carry different measurement meaning.
A customer identifier made of digits remains a nominal label, while response time is a continuous measurement. Ordinal ratings carry order but do not guarantee equal distance between categories. The management pipeline must preserve those distinctions when records are cleaned or joined. It must also document provenance, transformations, access and retention.
A polished table is not automatically trustworthy: missing populations, duplicate joins, outdated definitions or overly broad access can undermine the conclusion even when every cell is neatly formatted.
What this chapter covers
- 01
Numeric and non-numeric data
- 02
Discrete and continuous variables
- 03
Nominal and ordinal categories
- 04
Structured and unstructured records
- 05
Big-data management stages
- 06
Quality, privacy and security
Worked example · free
Audit a mixed service dataset
- +1Treat ticket ID as a nominal identifier despite its digits.
- +1Treat response minutes as continuous numeric data for time summaries.
- +1Treat issue category as nominal and compare frequencies.
- +1Treat satisfaction band as ordinal and preserve its order.
- +1Treat recordings as unstructured and code or transcribe them before analysis.
Key terms
- Discrete variable
- A numeric variable that takes countable values.
- Continuous variable
- A numeric variable that can take values throughout an interval.
- Nominal data
- Categories with no inherent ranking.
- Ordinal data
- Categories with a meaningful order but no guaranteed equal spacing.
- Data wrangling
- Cleaning, structuring and enriching raw data before analysis.
Data Types and Management FAQ
Why can a number still be nominal data?
Numbers can serve as identifiers or category codes without representing magnitude. Customer number 1047 is not larger in a meaningful business sense than customer number 1046. Arithmetic on those codes would add meaning the variable does not possess.
What is the difference between structured and unstructured data?
Structured data follows a predefined arrangement such as rows and columns. Unstructured data, including free text, images and recordings, lacks that fixed tabular form and usually requires feature extraction, transcription or coding before standard analysis.
How should I discuss a data-quality weakness?
Name the dimension, such as accuracy, completeness, consistency or timeliness, and explain its decision consequence. A missing customer segment can bias demand estimates, while inconsistent timestamps can reverse a service comparison. The consequence sets the repair priority.
Why must data provenance survive transformation?
Provenance records where a field came from, how its definition changed and which filters or joins shaped the final value. Without that trail, an analyst cannot tell whether an unexpected result is a real business signal or a pipeline artefact. It also lets reviewers reproduce, challenge and govern the evidence.
Assessment move
Build a classification table with columns for business meaning, numeric or non-numeric status, subtype, structure and valid operation. Add one privacy or quality risk for each field. When reviewing a pipeline, trace one record from generation to the final output and note each transformation. This exposes where definitions, missing values or access controls can change the result.
Recode the categories with different numbers and check whether their meaning changes; if it does not, the numbers are labels. For ordinal fields, practise selecting frequencies or a median without assuming equal gaps. Create a mixed record containing a table, free text and an image, then specify how each would be processed and joined.
Finish by writing the decision consequence of one quality weakness rather than calling the dataset messy. This turns classification into an analytical and governance choice. Build a second table that treats each transformation as a control point: source definition, cleaning rule, join key, access owner and validation check.
Introduce one duplicate, stale timestamp and missing category, then predict which summary or model output each fault would distort. Compare deletion, correction and explicit missingness as responses instead of assuming every imperfect record should be removed. Finally, state who may use the data and for which decision.
This rehearsal connects technical preparation with consent, purpose limitation and accountability rather than leaving governance as a separate final paragraph.
Working through Data Types and Management in BSAN2201? Sia is AskSia’s AI Business and Economics tutor — ask any BSAN2201 Data Types and Management question and get a clear, step-by-step explanation grounded in how BSAN2201 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.