Data Types
If you are wondering how to deal with data, how to explore it, and how to visualize it, well, this article is for you.
Data is everywhere, and data literacy or the ability to obtain meaningful information from data and thinking critically about it is essential nowadays. One way to improve your data literacy is to understand the basic concepts of exploratory data analysis or EDA for short. EDA is an excellent approach to summarize the main characteristics of a data set.
In this series of articles, you will be learning the main components of EDA, such as data types, data statistics, data relationships, and data visualization.
This first article is about the different types of data and how you can recognize them. The next article explores each data type using real data with various mathematical methods and visualization graphs. The data relationships will be discussed in the third article of this series.
Let’s get started
Data types
Understanding what kind of data you are dealing with is an essential phase to better extract insights from the data; since for each data type, you need to use the right mathematical techniques and visualization types.
There are basically two main types of data: Categorical and Continuous.
A. Categorical data
Data is labeled categorical if their values describe a quality, grade, level, or any other characteristics, such as happy/neutral/sad, low/high, dark/light. Categorical data can be Nominal or Ordinal.
1. Nominal data
Nominal data is a type of categorical data where the order doesn’t matter. Examples of nominal data:
- Male/Female.
- Countries’ names.
- Colors.
- …
To identify nominal data, ask yourself the following question: Will the meaning change if I change the variables’ order? If not, you have nominal data.
1.1. Statistical methods
Here are the main statistical methods used to explore nominal data:
- Frequency.
- Proportion.
- Percentage.
1.2. Visualization
Column/bar, pie, and donut charts are mainly used to visualize nominal data.
2. Ordinal data
Data is ordinal if it represents an ordered quality variable, and the number of units in each variable is not the same. Example of ordinal data:
Levels
- Low, medium, and high. Each level could have different units of measurement, such as the level of happiness, pain, etc.
Day’s times
- Morning: from sunrise to 11:59 am, it is around 6hours or 6 units of time.
- Afternoon: from 12:00 pm to 5:00 pm, it is 5 hours or 5 units of time.
- Evening: from 5:01 pm to 8:00 pm, it is 3 hours or 3 units of time.
- Night: from 8:01 pm to sunrise, it is around 10hours or 10 units of time.
Educational stages
- Preschool: from 3 to 5 years old, and it has one grade.
- Elementary school: from 5 to 14 years old, and it has 8 grades.
- High school: from 14 to 18 years old, and it has 3 grades.
- College: from 18 to 22 years old, and it has 4 to 5 years of education.
- Graduate school: ages vary.
2.2. Statistical methods
Here are the main statistical tools to explore ordinal data
- Frequency.
- Proportions.
- Percentages.
2.2. Visualization
Like the nominal data, column/bar, donut, and pie charts are often used to visualize ordinal data.
B. Continuous data
Data is labeled continuous if the values are measured, such as how many kgs of gold do I have? I have 30,5kg of gold; How big is the farm? The farm is 1km2 or 100 hectares (0.3861 square miles), etc. The data could be discrete (.., -1, 0, 1, 2, ..) or continuous (…, 0.123, 0,456, …).
Continuous data could be either interval or ratio.
1. Interval data
Interval data represents values where proportion doesn’t make sense, and the zero has no meaning. I can not say that a person with 160 IQ is twice as intelligent than a person with an IQ of 80, or a temperature of 20° is 20 times more than 0°.
Example of interval data:
- Intelligence quotient or IQ.
- Temperature.
- Time of the day, ex: 1 pm, 2 pm, 4 am, etc.
Remark
Interval data is often used in the social sciences.
1.1. Statistical methods
Here are the statistical tools to explore the interval data:
- Percentiles, median, interquartile range.
- Mean, median, or mode.
- Standard deviation, range, or IQR.
1.2. Visualization
Histogram and boxplot are the most used charts to visualize interval data since they are well suited to display main statistical elements such as percentiles, median, etc.
2. Ratio data
For ratio data, the zero has a meaning as a point of reference, and proportions make sense: I can say that 4kg is two times 2kg or 4 times 1kg, and the minimum weight is 0kg.
Example of ratio data:
- Distance: 5km, 1,84m, etc.
- Weight: 81.5kg.
- Days: 2 or 3 days.
- Times: 1 hour, 34min.
Remark
Notice that ratio data does not have negative values, as there is no negative height or weight. Ratio data is popular in physical sciences.
2.1. Statistical methods
Here are the statistical tools to explore the ratio data:
- Percentiles, median, interquartile range.
- Mean, median, or mode.
- Standard deviation, range, or IQR.
2.2. Visualization
Like the interval data, histogram and boxplot are also mainly used charts to visualize ratio data.
Summary
The best way to summarize this article is by setting up this table and this organigram as references:
Good to know
I have just scratched the surface with this basic data type introduction. In real-life, it could be challenging to identify which data type your data belongs to; since the process of data collection has a significant influence on the data types. But for now, let’s stick to the basics, and I will write another article about these real-life challenging data types.
Now, you have an idea about data types, and which statistical tools and graphs you can use to explore the data, let’s jump to the next article and see how you can apply this knowledge using examples.
