This article also undertakes the previous article “Data Visualization Using FineReport ”. Unlike the previous one, this time most of the advanced charts are more suitable for practical applications, and are developed by our developers for use in FineReport.

Sankey diagram – a visual analysis tool for energy, material composition, finance and other data

Sankey diagram is an energy split diagram commonly used for visual analysis of energy, material composition, and financial data. On the Sankey fan community website, there is a saying: “A Sankey diagram says more than 1000 pie charts”, which means that a Sanki diagram is richer than what is described in a thousand pie charts.

The application of Sankey is so niche. However, Sankey seems complicated, but it is very simple and efficient to use. For example, the energy flow diagram above, the width of the extended branch in the figure corresponds to the size of the data flow.

The characteristics of Sankey are as follows:
1. The initial flow rate and the end flow rate are the same, and the sum of all the main branch widths is equal to the sum of all the branched branch widths, maintaining the energy balance;
2. Internally, different lines represent different flow splits, and its width proportionally shows the flow occupied by this branch;
3. The different widths of the nodes represent the amount of traffic in a particular state.

The Sankey diagram consists of traffic, nodes, and edges, and is suitable for node data sets (optional) and edge datasets. Data weights are mapped to the width of nodes and edges. Sankey diagram needs to maintain energy conservation, not creating flow in the middle process, and the lost (loss) flow should flow to the node representing the loss.

The effect of the Sankey diagram plugin in the FineReport’s large screen:

Parallel Coordinates —“Dimensional Attacks” on Big Data

Parallel graphs are a visualization technique used to render multivariate, or high-latitude data, and are used to present relationships between multiple variables. Although a large number of segments are initially confusing, they are a very powerful tool for understanding multidimensional numerical data sets.

The method of describing parallel coordinates is generally to discuss high dimensional space, and how this technique arranges the axes in parallel instead of being orthogonal to one another. Below is the specific data sheet. The table details the models released from 1970 to 1982, including their mileage (gallon), number of cylinders, horsepower, weight, and the year in which they were produced.

Now let us imagine that each column maps to the vertical axis in the image above. Each data value ends somewhere along the line, scaling between the minimum at the bottom and the maximum at the top. However, purely collected points are not very useful, so points belonging to the same record (row) are connected to the line, which produces a feature blend of parallel lines.

By looking at this visualization, we can get a lot of information. The cylinder is prominent because it has only a few different values. The number of cylinders only is an integer, no more than eight here, so all rows must go through a small point. Such data and categorical data are generally not suitable for parallel coordinates. But if it is one or two, this is not a problem.

Between the miles of MPG and the cylinders that can be traveled per gallon of gasoline, you can see that eight-cylinder cars generally have lower mileage than six and four cylinders. If you follow the line to see how they intersect, you can see that many cross lines are signs of the inverse relationship. The graph shows the law: the more cylinders, the lower the mileage.

The correlation between cylinder and horsepower is more straightforward: the more cylinders, the more horsepower. Of course, there are also some cross lines , so more cylinders don’t always mean more energy, but the general trend is clearly there. The situation is similar between horsepower and weight: the greater the horsepower, the heavier the car, but of course there is some dispersion of value. Another exception is that a high horsepower eight-cylinder car is very light. Look carefully and find out the outliers.

Finally, the line between the weight and the year crosses a lot, which indicates that the car has become lighter over the years. You can also easily see that the annual axis only records a small number of different values, similar to a cylinder. Although this is a very simple example, it shows a typical structure in most datasets.

Parallel coordinates can be used to filter interactions. The main coordinate in the parallel coordinate system is called “brush”, and the image below should be obvious. To do this, let’s take a look at all the axes. Here, we brushed the range from 1980 to 1982 on the annual axis. The result is that part of the line is painted black and the rest is still grayed out.

In FineReport, our developer Da Jiangdong developed a combination of basic parallel graphs and maps or scatter matrices by encapsulating a well-known open source graph library.

The effect of the parallel coordinate graph plugin in FineReport:

Mosaic square chart, “don’t mess with it!”

Standard, non-uniform mosaics are used less in real life, mostly in the field of statistics, and are often used in some modules of SAS. Uniform mosaics are often used in life. A more classic example is the fare map between a subway station and a station.

Standard mosaics focus on a large number of data dimensions that are difficult for general users to understand intuitively. In general, it is recommended to use a uniform mosaic. For non-uniform mosaics, it can be broken into many different charts in most cases.

A mosaic with uniform coordinate axes is also a standard mosaic in the field of statistics. A uniform mosaic contains the following constituent elements: a uniform classification axis, a rectangular block with colors, and a legend.

From the data point of view, the uniform mosaic map and the heat map have very similar meanings on the continuous data, and the corresponding usage scenarios can be approximated. However, the heat map indicates that the color of the third dimension is linear, and the mosaic indicates that the color of the third dimension is classified. The standard heat map is subjected to a smoothing algorithm with no obvious boundaries, and the mosaic has clear boundaries.
Martin Theus wrote in his data analysis article “Understanding Region-Based Charts: Mosaics” that mosaics are Swiss army knives with categorical data. The bar chart stays in the limit of univariate, and the mosaic and its variants fully demonstrate the powerful visualization of multivariate classification data.

Diagram – Supported Forced Layout, Cartesian Coordinate System, Calendar Chart

Diagrams of relational classes usually display the relationship between data in a visual way. The relationship between data is represented by the nesting and position of the graph. It is usually used to indicate the order of the data, the parent-child relationship, and the correlation.
Common Sangiki and Wayne charts are counted as a form of representation.

Our developer smile integrates the well-known commercial open source chart library in China, which allows the diagram plug-in to support the layout of multiple relationship classes outside of the “circular layout”.such as the combination of relational data and Cartesian coordinate system, relational data. Forces the layout, as well as relational data combined with calendar diagrams, and supports different interactive animations and effects.

Map circle selection – use the mouse to draw a circle to see the corresponding data chart

The map circle selection is a special effect. The interpretation is to draw a circle with the mouse to see the corresponding data chart. The plug-in developer Xing Zhou has developed a combination of common scatter mark maps and bar charts for FineReport by packaging the famous domestic commercial chart library.

First of all, we only need to follow the traditional mark map to configure the name, value and location data of the map for the special selection. The plugin already has built-in coordinates for major cities in China, so when adding city markers, you don’t need to define the coordinate latitude and longitude positions separately.

The style panel supports the json map data built into fr, which is divided into world map, China map and provincial maps to facilitate the use in various scenarios.

The key point: When previewing, you can click the cross selection, free selection, keep selection, clear, etc. buttons in the upper right corner to mark the display in the map area.

At last
The above visual plug-ins are developed by the team members of the FanRuan Developer team, all from their own development or a part of the open source library, as a plug-in for the visual charts and animations of FineReport. FineReport is a commercial data reporting tool, related article:

Best reporting tool: a template is better than hundreds of Excel?

If you are our customer, you want to customize the FineReport plugin; or you are interested in the individual developer, you want to participate in the amateur project, you can participate in the development of the FanRuan.