Đại học Kinh tế Quốc dân – National Economics UniversityMICRO4ASIA
🏆 Arena
Sign inSign up
MICRO4ASIAThe six micro-credentials

Courses are delivered on the Limio learning platform. Limio

← Data Visualization in Business

You are viewing a public preview

Sign in to save your progress, do assignments, join discussions and get feedback.

Sign in
Module 1 — Overview of Data Visualization in Business

1.1 · Data and Data Analysis

Print / Save as PDF

In this lesson

Concepts and types of data; quantitative and qualitative data; structured, semi-structured and unstructured data; the relationship between data, information, knowledge and wisdom; data analysis in business contexts. The slides cover datasets, records and variables, independent and dependent variables, and the nominal, ordinal and quantitative (interval and ratio) data types with the operations each allows.

Learning outcomes

  • LO.1: Explain fundamental concepts of business data, data analysis, data visualization, chart types, dashboards and data storytelling.

The lecture slides for this lesson are below.

Study notes

By the end of this lesson you can:

  • Describe a dataset in terms of records and variables, and tell an independent variable from a dependent one.
  • Classify a field as nominal, ordinal, interval or ratio, and list the operations each type allows.
  • Distinguish quantitative from qualitative data, and structured, semi-structured and unstructured data.
  • Explain the data, information, knowledge and wisdom (DIKW) chain with a business example.
  • Explain why the type of a field limits which calculations and charts make sense.

Key idea: Before you draw any chart, know what kind of data each field holds. The data type decides which comparisons are meaningful, and therefore which charts and calculations are honest.

Datasets, records and variables

The lecture defines a typical dataset as n records (r1, r2, ..., rn). Each record holds m observations or variables (v1, v2, ..., vm), with m at least 1. In a spreadsheet, a record is a row and a variable is a column.

Variables can play two roles:

  • Independent variable: a variable that is not affected by another variable. Time in a time-series dataset is the classic example.
  • Dependent variable: a variable that changes with one or more associated independent variables, for example the temperature of a region over time.

In business charts the independent variable usually goes on the horizontal axis (Order Date) and the dependent variable on the vertical axis (Sales).

Four basic data types

The lecture groups data into nominal, ordinal and quantitative, and splits quantitative data into interval and ratio.

TypeWhat it isAllowed operationsExamples
Nominal (N)Labels or categories with no order=, ≠Fruit {apple, orange, pear}; colour {red, green, blue}
Ordinal (O)Values with a meaningful order=, ≠, <, >Film ratings G < PG < PG-13 < R; small < medium < large
Interval (Q)Ordered numbers with equal spacing but no absolute zero=, ≠, <, >, −Calendar dates, clock time, temperature in °C
Ratio (Q)Numbers with an absolute zero=, ≠, <, >, −, %Height, weight, price, quantity

Each type adds one operation to the type above it. With interval data you can measure a distance or span ("the order shipped four days after it was placed"), but you cannot say that one date is "twice" another. As the lecture notes, 14 September 2010 cannot be described as a ratio of 1 January 2011. With ratio data, statements such as "twice as much" or "a 20% share" make sense because zero really means "none".

Numbers are not automatically quantitative. A customer tier coded 1, 2, 3 is ordinal, and an order number is nominal even though it contains digits. A quick test: does adding or averaging these values mean anything?

Qualitative and quantitative; structured and unstructured

  • Quantitative data are measured as numbers (interval or ratio). Qualitative or categorical data describe qualities or groups (nominal or ordinal).
  • Structured data fit a fixed table schema with rows, columns and defined types, such as sales records in an ERP system.
  • Semi-structured data carry tags or keys but no rigid table, such as JSON, XML or email headers.
  • Unstructured data have no predefined model: customer reviews, images, audio and video.

Tableau works best with structured data, so other data are usually converted into tables first.

From data to wisdom

The DIKW pyramid describes how value is added as data move up the chain:

  • Data: raw facts, for example one order line with its quantity, discount and profit.
  • Information: data organised and given context, for example a summary showing that order lines with high discounts often have negative profit.
  • Knowledge: understanding why, for example "deep discounts on low-margin products push the price below cost".
  • Wisdom: judgement about what to do, for example "cap discounts on low-margin products and review the policy every quarter".

Data analysis in business is the work that moves an organisation up this pyramid: collecting, cleaning, summarising and interpreting data so that managers can decide. Visualization is one of the strongest tools for the step from data to information, because people see patterns in pictures much faster than in rows of numbers.

Worked example

The course practice data describe a fictitious office-supplies retailer. In the spirit of the classification exercise in the slides, classify some fields of the Orders and Customer tables.

  1. Order ID: a label that identifies an order. Equality is the only meaningful test, so it is nominal, even though it contains digits.
  2. Segment (Consumer, Corporate, Home Office) and Region (North, Central, South): categories with no natural order, so nominal.
  3. Ship Mode (Standard Class, Second Class, First Class, Same Day): if you rank the modes by delivery speed, the field becomes ordinal. Whether a field is treated as nominal or ordinal depends on whether the order matters for your question.
  4. Order Date: you can compute the gap between two dates, but "twice a date" is meaningless, so interval.
  5. Sales and Quantity: there is a true zero (no sales, no units) and "twice as many units" is meaningful, so ratio.
  6. Discount (a rate from 0 to 0.4): zero means no discount, so ratio.

So you can sum Sales, sort Ship Mode by speed, only count Segment values, and put Order Date on a time axis.

Common mistakes

  • Treating identifiers as numbers. Summing or averaging Row ID or a numeric customer code gives a meaningless result. Treat identifiers as nominal.
  • Losing the order of ordinal data. Sorting satisfaction levels alphabetically (High, Low, Medium, Very High) hides their natural order. Sort them by rank.
  • Ratio statements on interval data. Saying 20 °C is "twice as warm" as 10 °C is wrong because the Celsius scale has no absolute zero.
  • Confusing data with information. A table of raw rows is not yet an answer. Information needs a question and context.

Summary

  • A dataset is a set of records; each record has one or more variables, which may be independent or dependent.
  • Nominal, ordinal, interval and ratio data each allow more operations than the type before.
  • Numbers that act as codes or labels are not quantitative data.
  • The DIKW pyramid shows how analysis turns raw data into information, knowledge and decisions.
  • The data type of each field limits which calculations and charts are valid.

Flashcards · Data and Data Analysis

10 words
WordMeaningExample
  • Record

    One row of a dataset: a single observed item or event described by one or more variables.

    One order line in the Orders table.

  • Independent variable

    A variable that is not affected by other variables in the dataset; time is the classic example.

    Order Date on the horizontal axis of a sales trend.

  • Dependent variable

    A variable whose values change with one or more independent variables.

    Monthly Sales plotted against Order Date.

  • Nominal data

    Unordered categories or labels; only equal and not-equal comparisons are meaningful.

    Segment: Consumer, Corporate, Home Office.

  • Ordinal data

    Categories with a meaningful order, so greater-than and less-than comparisons are valid.

    Satisfaction: Low < Medium < High < Very High.

  • Interval data

    Ordered numbers with equal spacing but no absolute zero; differences are meaningful, ratios are not.

    Order Date: the gap between two dates in days.

  • Ratio data

    Numbers with an absolute zero, so differences and ratios such as 'twice as much' are meaningful.

    Quantity: 6 units is twice 3 units.

  • Semi-structured data

    Data organised with tags or keys but without a fixed table schema.

    A JSON file of online orders.

  • Unstructured data

    Data with no predefined model, which must be processed before it can be charted.

    Free-text customer reviews or product photos.

  • DIKW pyramid

    The chain data, information, knowledge, wisdom, showing how analysis adds context and judgement to raw facts.

    Raw order lines become a discount policy decision.