Contents show

Getting a hang on statistical data types is critical in data analytics, particularly while performing exploratory data analysis (EDA) and machine learning (ML) projects. This understanding helps apply the most accurate statistical methods and validate hypotheses during projects. Whether you are a researcher, student, business personnel or a data analytics enthusiast, having a great command over various data types in statistics ensures your results are accurate, while also significantly streamlining the analysis process.

In this article, we will have a comprehensive overview of different statistical data types.

Image Source: Leverage Edu

What is Data?

Data is a collection of characters, numbers, images or other values that convey a specific information that can be translated into a form feasible for processing or movement. In statistics, you systematically collect and categorise some facts and quantitative information and analyse it to extract actionable insight. You have a range of options and methods to collect statistical data, for example, observation, surveys, open-ended questionnaires, etc. The nature and specifications of your research determines which method to choose to collect this data.

What is Statistical Data

Statistical data refers to the collection of quantifiable facts and information systematically collected, organised, and analysed. You can collect statistical data with a range of techniques – surveys, interviews, telephone interviews, open-/closed-ended questionnaires, etc.

Based on the nature of data and how its collected, statistical data can be classified into two forms:

Qualitative data and
quantitative data.

Qualitative Data

Qualitative data, better known as categorical data, is non-numerical, and descriptive information often used to define and categorise the characteristics or attributes of an individual or event. Some common methods to qualitative data collection are interviews, open-ended surveys and questionnaires, observation, content analysis, etc.

Identifying trends and patterns through coding, thematic analysis, content analysis is critical for seamless analysis of qualitative data.

For example, assume you are a restaurant owner studying your customer feedback. In this case, your qualitative data would include customer feedback about the food, service or ambience of the place that you would categorise into themes.

Nominal Data

Nominal data is a collection of some distinct, individual elements that you can classify or label into unique and mutually exclusive categories within a variable. Since nominal data are some discrete units you cannot meaningfully measure or order these categories. With nominal data, you can label variables but cannot quantify them.

Some examples of nominal data are:

Name
Nationality
Gender
Physical attributes, such as eye colour
Zip code, etc.

In data science, hot encoding is used to give nominal data relevant numeric equivalence. Nominal data is used in descriptive statistics that includes where information can be collected through:

Frequencies: In nominal data, frequency refers to the number of times a specific category appears in the data set.
Proportion: A proportion in nominal data is the ratio of the frequency of a specific category to the total number of observations in the data set.
Visualization Methods: Pie charts and bar charts are used to visualise nominal data.

Key features of nominal data:

a) No order: There is no specific order to follow in the nominal data categories. For example, if you are categorising pets into categories like “dog”, “cat” or “rabbit”, there is no inherent order here that can impact the outcome.

b) Distinct labels: You need to label each category of nominal data you have gathered to make them easily identifiable and classifiable.

Ordinal Data

Ordinal data represents data that can be ranked, ordered or categorised. However, ordinal data goes beyond categorisation like nominal data as it has the “order” element added. In this data type, you can order data in a sequence or rank but the distance between two categories are not always equidistant. Like nominal data, ordinal data has no meaningful zero.

Examples of ordinal data:

Educational qualification
Opinion (neutral, agreed, mostly agreed, disagreed, mostly disagreed)
Socioeconomic condition (low income, working class, middle income, wealthy)
Client contentment score
Likert scale response
Exam score classification

Key characteristics of Ordinal Data include:

a) Order: Ordinal data is a type of categorical data in which the categories can be logically ordered or ranked. For example, you can categorise customer satisfaction ratings and order them in ranking like satisfied, very satisfied, dissatisfied, very dissatisfied and more.

b) Non-equidistant intervals: To reiterate, it’s not necessary for categories in ordinal data to be equidistant.

Label encoding is used to translate ordinal data into numeric equivalence.

You can implement a range of descriptive statistical techniques – proportions, central points, percentages, frequencies, median, mode, interquartile range, etc., – on ordinal data. The descriptive statistics that you can do with ordinal data include frequencies, proportions, percentages, central points, percentiles, median, mode, and the interquartile range.

To visualise ordinal data, you can use pie charts and bar charts.

Numerical or Quantitative Data Types

Quantitative data is measurable, or countable numeric values. You can apply statistical and mathematical analysis on quantitative data to test hypotheses, or compare and predict events with numerical values – precisely. To collect quantitative data, you can use polls, close-ended questionnaires, telephone interviews, longitudinal studies, etc.

Different types of statistical data in quantitative studies include:

Discrete Data

Discrete data involves values that are distinct and separate. This type of data takes specific, countable, finite and fixed values and cannot be measured in the way continuous data can be. It can be itemised and listed one-by-one. For example, the number of students in a class or the number of employees in an organisation are discrete data because they can be counted as whole numbers, and cannot be divided into smaller, meaningful parts.

Examples of discrete data:

Number of computers in a lab
Number of employees in an office
Number of students in a class
Number of clients
Number of clicks on a website

Some features of discrete data include:

a) Countable values: Discrete data are countable, listable and distinct values.

b) Intervals between values: You can see noticeable and distinct intervals between data points. It means that there exists no data within these intervals and two between two data points.

Continuous Data

Unlike discrete data, continuous data can be of any value within a specific range and can be collected through measurements, not counting. Within a specific range, it can be of infinite value expressed in decimal or fractions. It means that you can break down continuous data into finer levels. For example, the weight, or height of a person, humidity in the weather, temperature of a room, etc. are continuous data as they can take any value and are fragmented into smaller increments.

Examples of continuous data:

Height
Weight
Distance
Temperature
Time

Continuous Data is characterised by two key features:

a) Infinite possible values: Continuous data can have infinite values meaning that you can have an infinite number of data between two data points.

b) No gaps or discontinuities: There are no discontinuities, intervals or categories between values in continuous data.

Interval data

The measurements in interval data can be ordered and in two specific measurements, you will have the same difference. With “interval”, it can be understood that there is an equal difference between two data points throughout the scale.

For example, if you’re measuring temperature in degrees Celsius, the interval or difference between 10 and 20 degrees is the same as the interval between 20 and 30 degrees, which is a difference of 10 degrees. An example of interval data is the measurement of temperature. If you have two measurements at 30 degrees and 40 degrees, the difference between these two points is the same, namely 10 degrees. However, in the degree Celsius or Fahrenheit scale of temperature, there is no absolute zero point exhibiting an absence of temperature.

This lack of a true zero point in the variable scale is a key characteristic of interval data, and a differentiator of interval data from ratio data where true zero exists. In interval data, the absence of a “true zero” means that there is no point at which the feature of the variable being measured does not exist. Therefore, it won’t be accurate to make statements about the ratios of two values of the variable under experiment.

For example, you can’t say 40 degrees is twice as hot as 20 degrees. It’s because in the temperature scale, zero indicates another value, not non-existent temperature. Therefore, ratio comparison doesn’t hold good for interval data as they do for ratio data where true zero exists. You have a range of statistical techniques to apply on interval data. However, due to lacking true zero in interval data, division and multiplication cannot be applied on it.

Ratio Data

In ratio data, you can categorise and order the variable under study and get meaningful insights by finding the interval between two data points. It means that ratio data is the same as interval data with the presence of true zero.

The presence of true zero point in the ratio data allows for the effortless implementation of a range of statistical analysis, for example, variability, central tendency, correlation, etc., – and make meaningful statements about the ratio of two measurements. For instance, age is a good example of ratio data. It starts from the zero point and can be compared with other values, such as, a 30-years man is two times older than a 15-years old person.

Author
Recent Posts

George Wilson

Data Science and Business Intelligence Strategist at Symbolic Data

George Wilson is the Lead Editor at Symbolic Data, where he spearheads the editorial direction and content strategy. With over a decade of experience in business intelligence and data management, George has established himself as a thought leader in the field. His expertise lies in translating complex data concepts into actionable insights for business executives and CEOs.