Read on Twitter
Heads up: A thread explaining the basics of analytics coming this Sunday.
I will give examples of charts you would typically use for simple data analysis (with types).
I will give examples of charts you would typically use for simple data analysis (with types).
1/
If you ever wanted to learn (in a simple way), why one chart is used over the other in data analytics, this multi-series thread is for you.
Let's begin:
If you ever wanted to learn (in a simple way), why one chart is used over the other in data analytics, this multi-series thread is for you.
Let's begin:
2/
If you simply want to show the 'change of a value over time', use:
- A simple numeric display to show the latest value
- Show change % vs. previous period
If you want the same to be more explicit:
- Simple columns to plot previous and current
- Optionally, put change %
If you simply want to show the 'change of a value over time', use:
- A simple numeric display to show the latest value
- Show change % vs. previous period
If you want the same to be more explicit:
- Simple columns to plot previous and current
- Optionally, put change %
3/
If you want to show the 'trend' of value over time, use:
- Line (if just the trend is important to display)
- Column (if individual values are also important to navigate)
(There are more reasons to choose between line and column to plot time-series; more on that later)
If you want to show the 'trend' of value over time, use:
- Line (if just the trend is important to display)
- Column (if individual values are also important to navigate)
(There are more reasons to choose between line and column to plot time-series; more on that later)
4/
If you want to show the breakdown of a value:
(We call it 'to segment' in some context)
- Use a table layout for a simple breakdown
- Use a bar chart if you want to compare the objects in the breakdown as well
If you want to show the breakdown of a value:
(We call it 'to segment' in some context)
- Use a table layout for a simple breakdown
- Use a bar chart if you want to compare the objects in the breakdown as well
5/
When there are many objects, it's easier to 'highlight' one object of interest to show how it compares against others. We often call it 'benchmarking'.
Simply highlight the object with a different color/ formating.
In this e.g., 'where does Texas stand against other states'
When there are many objects, it's easier to 'highlight' one object of interest to show how it compares against others. We often call it 'benchmarking'.
Simply highlight the object with a different color/ formating.
In this e.g., 'where does Texas stand against other states'
6/
If you want to see the contribution of individual objects to the total value, use a stacked bar or simple bar.
This is the same as the 'contribution of parts of the whole'.
You may add contribution % as well.
If you want to see the contribution of individual objects to the total value, use a stacked bar or simple bar.
This is the same as the 'contribution of parts of the whole'.
You may add contribution % as well.
7/
For '6', we don't prefer the 'pie' chart since it performed poorly in our UTs (more on that later).
We use about 25-30 analysis types to determine which chart type is 'most relevant' for a user.
Next week, I will post part 2 of this series.
Stay safe, happy weekend.
For '6', we don't prefer the 'pie' chart since it performed poorly in our UTs (more on that later).
We use about 25-30 analysis types to determine which chart type is 'most relevant' for a user.
Next week, I will post part 2 of this series.
Stay safe, happy weekend.
