How to make a diagram - a formalized algorithm
Initially there was data. You have already analyzed them, found interesting dependencies or, on the contrary, the absence of dependencies. Perhaps even built an intermediate chart to identify these dependencies. The main thing is not to stop, because what happened by itself does not always have the best.
The following is a simple formalized algorithm for how to illustrate data with a diagram. In short, he is:
Idea → Comparison Type → Chart Type
')
The idea is the very essence of what we are going to tell the user with the help of a diagram. It is born either “itself”, as a result of the discovery of dependence, or “from the mind”, when you want to confirm your thoughts, ideas, your way of life during the last couple of years, and maybe even horror! - inspire something to the target audience; this is how anyone succeeds :-)
Her idea, we put in the title (important point).
A “finance allocation scheme” is not an idea. But the headline seems to be pulling ... "The dependence of work experience and efficiency among employees" is also. Not an idea? Get out! Looking for ideas. And so here they are ...
"The benefits are distributed evenly" - and the Communists are satisfied. But because the idea! And do not think.
“The efficiency of the work of our employees does not depend on the work experience” - well, there can already be a backlash.
"The company's share is growing steadily . " The investor, satisfied, rubs the portfolio.
Any idea can be expressed through five types of comparisons: component-wise, positional, temporal, frequency, and correlation. Thanks to Jean Zelazny for this sacred knowledge - I myself did not realize that this area can at least try to formalize and classify. So, looking at the formulated idea , we select the appropriate type of comparison.
1) exploded comparison . Shows the percentage of each component as a percentage of the whole.
- Two customers brought us half the profit.
- The boys in our faculty this year, only 90%.
- Sales of security patches to our system accounted for the largest share in the total sales of the company.
We look at the idea, if it contains the keywords “share”, “percent of the whole”, “made up X percent”, then we are here.
2) Positional . Shows how objects relate to each other.
- Staff turnover in six divisions is about the same.
- The director of the company does 24% less headshots at the CS than the average employee.
- Sales of left shoes exceeded sales of the right.
Keywords: “more than”, “less than”, “equal”.
3) Temporary . Shows how the value changes over time.
- All year the country's economy has steadily flew to hell.
- The dollar has already hesitated to grow.
- Sales in January grew steadily.
Keywords: “change”, “grow”, “decline”, “fluctuate”.
4) Frequency . Shows how many objects fall into certain successive areas of numeric values.
- 20% of employees perform 80% of the work.
- Most transactions were concluded on from X to Y rubles.
- The age structure of kindergartens differs from that in high school.
- In August, the largest share of stupidity was done with the expression “Serious”.
Key words: “in range”, “concentration”, “frequency”, “distribution”.
5) Correlation . Shows the relationship between two variables.
- The results of the May sales demonstrate the lack of connection between sales and experience of sellers.
- The growth of the state of nannies contributes to the development of monocular vision in the child.
- Minimization of effort by the fisherman does not lead to an increase in the catch.
Key words: “increases / decreases / changes with (in case)”, “connection”. Or, on the contrary, “does not change”, “does not increase”.
(To cope with this classifier required some effort from me. Or maybe it wasn’t enough, ha-ha.)
There are also five types of diagrams:
- circular
- ruled
- bar chart,
- schedule
- point.
We choose according to the scheme:
( copy of the picture )
Consider limiting the number of sectors in a pie chart according to the maximum number of objects that fit in the brain's cache: no more than 5-7. Minorities are then merged.
Frequency comparison . Use a "discrete" histogram if the values ​​are no more than 5-7, otherwise - the graph.
Correlation comparison . If there is little data, try a ruled one, if there is a lot of dotted data.
There is also a bunch of chart subspecies - combined, with accumulation, normalized, stepped, range, deviations, blown up - but not this time. And so on.
The following is a simple formalized algorithm for how to illustrate data with a diagram. In short, he is:
Idea → Comparison Type → Chart Type
')
Looking for an idea
The idea is the very essence of what we are going to tell the user with the help of a diagram. It is born either “itself”, as a result of the discovery of dependence, or “from the mind”, when you want to confirm your thoughts, ideas, your way of life during the last couple of years, and maybe even horror! - inspire something to the target audience; this is how anyone succeeds :-)
Her idea, we put in the title (important point).
A “finance allocation scheme” is not an idea. But the headline seems to be pulling ... "The dependence of work experience and efficiency among employees" is also. Not an idea? Get out! Looking for ideas. And so here they are ...
"The benefits are distributed evenly" - and the Communists are satisfied. But because the idea! And do not think.
“The efficiency of the work of our employees does not depend on the work experience” - well, there can already be a backlash.
"The company's share is growing steadily . " The investor, satisfied, rubs the portfolio.
The idea may not be born, effort is needed here, but without it, this algorithm is simply not applicable further. Then draw an illustration according to the usual intuitive scheme: there will be an ordinary descriptive title (“this is the house, son”) and a presentation in the spirit of “I've drawn something here, and you think yourself”. In this case, everyone in the audience will have to perform the data analysis work. If this is your goal, feel free to go this way.
Choose a type of comparison
Any idea can be expressed through five types of comparisons: component-wise, positional, temporal, frequency, and correlation. Thanks to Jean Zelazny for this sacred knowledge - I myself did not realize that this area can at least try to formalize and classify. So, looking at the formulated idea , we select the appropriate type of comparison.
1) exploded comparison . Shows the percentage of each component as a percentage of the whole.
- Two customers brought us half the profit.
- The boys in our faculty this year, only 90%.
- Sales of security patches to our system accounted for the largest share in the total sales of the company.
We look at the idea, if it contains the keywords “share”, “percent of the whole”, “made up X percent”, then we are here.
2) Positional . Shows how objects relate to each other.
- Staff turnover in six divisions is about the same.
- The director of the company does 24% less headshots at the CS than the average employee.
- Sales of left shoes exceeded sales of the right.
Keywords: “more than”, “less than”, “equal”.
3) Temporary . Shows how the value changes over time.
- All year the country's economy has steadily flew to hell.
- The dollar has already hesitated to grow.
- Sales in January grew steadily.
Keywords: “change”, “grow”, “decline”, “fluctuate”.
4) Frequency . Shows how many objects fall into certain successive areas of numeric values.
- 20% of employees perform 80% of the work.
- Most transactions were concluded on from X to Y rubles.
- The age structure of kindergartens differs from that in high school.
- In August, the largest share of stupidity was done with the expression “Serious”.
Key words: “in range”, “concentration”, “frequency”, “distribution”.
5) Correlation . Shows the relationship between two variables.
- The results of the May sales demonstrate the lack of connection between sales and experience of sellers.
- The growth of the state of nannies contributes to the development of monocular vision in the child.
- Minimization of effort by the fisherman does not lead to an increase in the catch.
Key words: “increases / decreases / changes with (in case)”, “connection”. Or, on the contrary, “does not change”, “does not increase”.
(To cope with this classifier required some effort from me. Or maybe it wasn’t enough, ha-ha.)
Choose a chart type
There are also five types of diagrams:
- circular
- ruled
- bar chart,
- schedule
- point.
We choose according to the scheme:
( copy of the picture )
Consider limiting the number of sectors in a pie chart according to the maximum number of objects that fit in the brain's cache: no more than 5-7. Minorities are then merged.
Frequency comparison . Use a "discrete" histogram if the values ​​are no more than 5-7, otherwise - the graph.
Correlation comparison . If there is little data, try a ruled one, if there is a lot of dotted data.
There is also a bunch of chart subspecies - combined, with accumulation, normalized, stepped, range, deviations, blown up - but not this time. And so on.
Source: https://habr.com/ru/post/51671/
All Articles