_edited_edited.png)
Through this article I am sharing my own personal experience with data analytics.
I have gained knowledge about how to analyze data and acquired proficiency in creating charts using tools such as Tableau, Power BI, and Python. However, I have encountered difficulties in determining the most suitable chart for my analysis. If you find yourself in a similar situation, this article is tailored for you.
Having complex or colorful charts doesn’t necessarily make your analysis good; it’s important to keep the charts simple, uncluttered, and use mild colors, while highlighting the key business numbers.
And we all know that visual cues help a lot to grab people’s attention and to leave a lasting impression on people’s mind in any presentation. Remember, data is converted into valuable information only if you know how to visualize it and get business insight out of it properly.
So thus, choosing the right chart is one of the most important aspects of data analysis.
Few things to figure out at first before plotting the chart,
· think about the message you want to share with your audience,
· ask yourself what specific insights or information you want to highlight,
· how many variables you want to show,
· how many data points you want to display, and
· how you want to scale your axis
Use the categories in the below flow diagram, to decide which data relationship is more important in your story, then look at the different types of charts within the category to form initial ideas about what might work best. This list is not very exhaustive, but I am sure it will be a very useful starting point for making informative and meaningful data visualizations.
let’s distinguish four main chart types:
Comparison
Comparison charts are used to compare items or show the variable differences over time.
Correlation
These charts are used to show a connection or correlation between two or more variables.
Composition
Composition charts are used to display parts of a whole and change over time. These charts provide a visual representation of how different components contribute to the overall composition and how their proportions evolve over a specified period.
Distribution
These charts are used to show how variables are distributed over time, helping identify outliers and trends. By illustrating the spread and pattern of data points, distribution charts enable analysts to find out patterns, anomalies, and the overall shape of the data distribution.
The dataset being used below is the Superstore Dataset available on Tableau Community or Kaggle website
Let’s go through the brief explanation on each category of charts.
Under the Comparison category, we have 3 subdivisions based on the variable we want to display. They are Magnitude, Ranking and Change over time sections.
Magnitude — (Bar/Column Chart, Paired Column, Proportional Stacked Bar)
Show size comparisons. These can be relative (just being able to see larger/bigger).
To show the count (number of) of the variable in each category.
Bar/Column chart
The standard way to compare the size of things. Must always start at 0 on the axis. Good when the data are not time series and labels have long category names.
Best Practice,
· Start the x-axis at 0 for bar and y-axis at 0 for column charts respectively.
· Label the axes
· Put value labels on bar
· Avoid using too many colors, creating the “rainbow effect”, instead use a single color, or varying shades of the same color.
· You can highlight one bar if that’s the message you want to get across.
Paired Column
As per standard column but allows for multiple series. Can become tricky to read with more than 2 series.
Below chart indicates the comparison of sales of each segment for each year (time series 2020 and 2021).
Proportional Stacked Bar
A good way of showing the size and proportion of data at the same time — if the data is not too complicated.
Ranking (Ordered Bar Chart, Ordered Column Chart, Lollipop Chart)
Use where an item’s position in an ordered list is more important than its absolute or relative value. When we sort the regular bar or column chart, we will get ordered bar, ordered column and lollipop chart.
Ordered Bar Chart — Showing sales for each ship mode. (Standard Class — Highest Sales)
Ordered Column Chart — Showing sales for each subcategory. (Standard Class — Highest Sales)
Lollipop Chart — Showing sales/Profit for each subcategory. (Red colored dots — Negative Profit).
Change over time(Line, Column, Area, Line+Column, Calendar Heatmap)
Give emphasis to changing trends. These can be short (intra-day) movements or extended series traversing decades or centuries: Choosing the correct period is important to provide suitable context for the reader.
Line
The standard way to show a changing time series. It reveals trends or changes over time. It can be used to show relationships within a continuous data set.
Column
Columns work well for showing change over time — but usually best with only one series of data at a time.
Line + Column
A good way of showing the relationship between an amount (columns) and a rate (line).
Area Chart
Area charts are a lot like line charts, with a few subtle differences. They can both show change over time, overall trends, and continuity across a dataset. Line charts are good for showing the net change in population over time, while area charts are good for showing the total population over time. One major drawback with this chart,
It can be good at showing changes to total, but seeing changes in components can be very difficult.
Calendar Heatmap
A great way of showing temporal patterns (daily, weekly, monthly).
Below image shows the monthly sales data.
Correlation(Scatter plot, Bubble Plot and Line + Column chart)
Show the relationship between two or more variables. Please note that unless explicitly stated otherwise, readers often assume a causal relationship between the relationships presented (i.e. one causing the other). It is important to be mindful of this when presenting information.
Scatterplot
The standard way to show the relationship between two continuous variables, each of which has its own axis. There is a positive trend line for the two variables used below.
Bubble Chart
Like a scatterplot but adds additional detail by sizing the circles according to a third variable.
Composition(also mentioned as Part-to-Whole)
(Stacked Column, Proportional Stacked Bar, Pie Chart, Donut Chart and Tree map)
Show how a single entity can be broken down into its component elements. If the reader’s interest is solely in the size of the components, consider a magnitude-type chart instead.
Stacked Column
A simple way of showing part-to-whole relationships but can be difficult to read with more than a few components.
Proportional Stacked Bar
A good way of showing the size and proportion of data at the same time — as long as the data are not too complicated. (please see ab
.
Pie Chart
A common way of showing part-to-whole data — but be aware that it’s difficult to accurately compare the size of the segments.
Donut Chart
Similar to a pie chart — but the center can be a good way of making space to include more information about the data (e.g., total).
Distribution (Histogram, Dot Plot, Box Plot, Barcode Plot, Dot Strip Plot)
Show values in a dataset and how often they occur. The shape (or ‘skew’) of a distribution can be a memorable way of highlighting the lack of uniformity or equality in the data.
Below is the dot plot representing sales amount of each sub-category between 2018 and 2021. (as per dataset used here)
Spatial charts
It can precise locations and geographical patterns in your data. Showing the airport terminals with the most foot traffic or a map of all sales across the country are examples of spatial maps.
Spatial maps include filled maps, point distribution maps, symbol maps, and density maps.
Conclusion
Effective data visualization is a key element in conveying meaningful insights to your audience. The choice of the right chart type plays a pivotal role in ensuring that your message is clear, accurate, and easily comprehensible.
Note: Used Tableau Desktop 2024.1 to create all visualizations used in this article.
Reference:
Comments