[Data] group of scientists studying data visualization (II)

This series of articles has two main purposes:

From the point of view of data visualization, to sum up now both data visualization or report development, BI development tools and ideas
Learn from east lay the wood study group to study the idea of some manufacturers work so that they think are more comprehensive and more mature, complete documentation is also more comprehensive, beautiful.

Learning Content

"Time series data" and "the proportion of data" Visualization

Understand the characteristics of time series data, different types of master timing chart for each type of data;
To know the percentage Data visualization purposes, the feature data set to learn to choose the desired table;
Practice: 4 from a given data set selected two, and charts to select the appropriate visual rendering

Knowledge summary

Time series data:

Discrete points in time (fewer cycles): Histogram, parallel bar, stacked bar charts, histograms and other percentages stack
- Highlights: the data do not constitute too much
- Category accounting change: using a percentage stacked bar chart
- Contrasting: the number of not too much data
Continuous time points (multiple cycles): Line charts, area charts, stacked area chart, waterfall diagram (ladder diagram), fitting curve like FIG.
- FIG dotted line: no more than 12 data items
- Line chart: When the data set more items, when greater than 12, in dotted line in FIG., The entire line will point very dense, impact upon the trend of the data, this time using the line graph is a good choice
- Graph: neighboring node connections smoother, better reflect trends
- FIG step: a time two adjacent nodes, the node data with respect to a change in the lift of the previous node data commonly used in commodity price changes, fluctuations in the stock price, the rate change in the scene.
  - Node previous time value;
  - The current time value of the node;
  - Representing a difference between the current node before the node;

The proportion of data:

Free time attributes: a pie chart, ring chart, pie chart nested, rectangular tree;
- Pie
  
  When less data items constitute an integral, pie chart is a good choice. Recommended maximum number of sectors between 5 to 7. When the data item exceeds a certain amount, can be accounted for in accordance with, the rankings last several homing "Other"
- Donut
  
  Pie sector angle through the demapping corresponding to each classification item value "FIG annular" is the value measured by the length of each arc.
- Rectangular tree
  
  Rectangular tree, is a method based on the visual area. Outer rectangle represents the parent class, inner rectangle represents the sub-categories. Compared to other types of data represented by the ratio, a rectangular display is more suitable for tree having a tree data structure.
Time attribute comprising: Percentage stacked column, the percentage of the stacked area chart;

finish homework

Scene 1: A company's 2018 electricity supplier with annual sales trend

Data Scenario 1 provides a summary of data from the point of view of data granularity, information expressed in each row of data is not clear, it is necessary to do the pre-processing of data, each row of data will become a monthly sales data , measure It is a sales dimension is a commodity classification.

End-use stacked bar chart shows the changes in gross sales, with a different color-coded classification of goods (gross monthly trend line is not successfully added on)

Scenario 2: change in a stock price in October 2019

This data directly ladder diagram (FIG candles or waterfall plot), note that the use of color and size tableau options to control the color and pattern of the column size

Scene 3: A day courier company mailing list Lanshou aging distribution

FIG rectangular tree directly

Scenario 4: Mainstream Express carrier volume flow distribution

Scene 4 complex, strictly speaking need to use two figures, but if you do not look at the proportion of the number of long words is a drawing can be done