Before I start, I'd just like to say that if you have any questions,
feel free to ask for clarifications. I am willing to work until you
are completely satisfied with the answer.
The major types of graphs are as follows (I will describe each of them
in detail after):
Graphs are visual representations of data. They can take on many
forms, primarily the ones listed above. They allow people to quickly
absorb information, observe trends and to easily interpolate and
extrapolate data. They are much easier to understand then a large
table of numbers and if well constructed, should provide the same
amount of information as the table. In presentations, it is usually
preferable to use graphs as they convey your point quickly and without
the need to understand what each and every number in a table means.
Column graphs typically have two axis, an x-axis (horizontal) and
y-axis (vertical). The x-axis is usually labelled with the categories
being compared. The y-axis is usually labelled with the frequency, or
value of each category. In bar graphs, it is the other way around
(thus, the bars are horizontal,
However, the terms bar and column are usually used interchangeably. In
both bar and column graphs, the greater the height, the greater the
Bar graphs are used to highlight separate quantities, especially the
differences between these quantities. They are extremely useful for
comparing quantities in different categories, and can be used to
describe the relationship of several variables at once. The data
typically being represented is the number of "occurrences" measured in
different categories of data. They are used in almost every field.
- Excellent for data comparison (esp. vertical bar graphs)
- Labelling for clarification possible with horizontal bar graphs
- Clearly show error values in the data
- Usually simple to read and understand
- Can be tempting to compare too many things, graph becomes convoluted
and difficult to understand
- Limited space for labelling with vertical bar graphs
3-D Bar Graphs (http://www.jpowered.com/bar_graph/vbargraph/): Used to
add visual flair to the graphs, but tends to make the graph harder to
read. They are used in almost every field.
3-Axis Bar Graphs: Includes an extra z-axis. Used for comparison of
categories with two changing properties. Can be extremely confusing
for the viewer. They are used primarily in business when demonstrating
for example, the success of a product as a function of two properties.
Stacked Bar Graphs (http://www.statcan.ca/english/edu/power/ch9/images/bar7.gif):
Used to show segments of totals. Usually used for percentages.
However, they can misrepresent data and confuse the viewer. There are
much better ways of presenting segment data. Should be avoided if
possible. They are used primarily in business to show product data.
Line/Dot Bar Graphs (http://www.statcan.ca/english/edu/power/ch9/images/bar8.gif):
Same function as a bar graph. The advantage of this type of graph is
that it is graphically very simple, and should be used if there is a
lot of things being compared so the focus is on the data, not the
graph itself. They are used in almost every field.
Very similar to bar graphs, except there is no space between bars.
Used to show the frequency distribution of a continuous variable (i.e.
the heights of students in a class). They are used a lot in
statistical work and demographics.
Used to represent the same information as a bar graph, except in a
more visially appealing way. Pictures are used to represent values.
The legend involves defining what each symbol represents. Must be done
visually accurate so not to misrepresent data. Symbols chosen are
usually pertinent to the data itself. They are used in almost every
One of the most popular types of graphs, line graphs have two axis.
The horizontal (x-axis) is for the independent variable, and the
vertical axis (y-axis) is for the dependent variable. Points on the
graph are connected by lines, hence the name.
Line graphs are typically used to show how a value changes over time,
though the independent variable can really be anything. In a generaly
sense, they are used to show how one value changes as another changes
uniformly or incrementaly. They are used in almost every field.
- Show specific values of data well
- Reveal trends and relationships between data
- Compare trends in different groups of a variable
- Clearly show error values in the data
- Usually simple to read and understand
- Inconsistent scales/different scale start points can distort the
data so it is interpreted incorrectly (biased)
- Multiple lines on the graph, especially unrelated can be confusing
- Labelling tends to convolute graphs, difficult to discern exact values for data
Area Line Graph (http://www.jpowered.com/area_graph/Examples/appletexample7.htm):
These graphs the same information as a normal line graph, except the
area underneath them is shaded. They are used to visually emphasize
the difference between the lines. Notice in the example that Product Y
and Product X approach the same value. The greater the amount of a
colour, the greater the difference between two lines. Typically used
when the lines represent very similar things, such as products.
Surface Graph (http://wwwslap.cern.ch/comp/doc/NExS/html/node265.html):
A surface graph enables you to show a trend in data across two
dimensions in a continuous curve. Modelling a value as a function of
its horizontal and vertical location is a primary example of its use.
Pie graphs, in their simplest form, are circles subdivided into
different coloured regions. The greater the slices area, the greater
the categories' value.
Pie charts are typically used to summarize categorical data, or even
more often, percentile data. The components have to add up to make a
"whole" of sorts or else the graph becomes meaningless (ex. student
population, market segment, etc...). A chunk may be seperated from the
rest of the pie to indicate its significance. They are used in almost
- Provides an excellent visual concept of a whole
- Clear comparison of different components
- Highlight information by visual seperation of a segment
= Easy to label, lots of space
- Comparing pie graphs is very difficult as pie graphs indicate
components' sizes relative to each other, not to some absolute value
- Too many segments is difficult to read, hard to label; better off
using a bar graph
- Difficult to understand without labels (especially with similarly sized segments)
- Hard to illustrate error values
3D Pie Graphs (http://www.jpowered.com/pie_chart/): 3D pie graphs add
a level of "sophistication" to the chart (though, nowadays, it's so
commonplace that it is more often misused). The added perspective
however, tends to artifically inflate the size of the segments on the
top and bottom of the circle and artificially minimize the segments on
the side. The severity of this distortion depends on how heavily the
3D is applied. They are used in almost every field.
Donut Graph (http://www.visualmining.com/examples/nc4styles/imgexamples/images/dialdonut.png):
A visual variation on a pie graph, conveys the same information.
Information can be placed in the middle of the donut to draw attention
to it. They are used a lot in advertisement.
Scatter Plot Graph
Scatter plots are generally composed of 2-axis (I've yet to see a
3-axis scatter plot). Both the x-axis and y-axis represent a range of
values. Where the axis intersect is always (or should be), (0, 0).
Data points are plotted according to their x/y-values, but they are
Scatter plots illustrate paired data, that is, information regarding
two related variables. It is particularly useful when one the
variables (represented by the y-axis) is dependent on the other
(represented by the x-axis). The resulting pattern (after all the
points have been plotted) will indicate the strength of the
correlation between two variables. Mathematical methods have been
developed to create lines of "best fit" which are essentially
equations which model the relationship between the two variables.
These relationships can be used to project how the variables would
react if one were to be increased beyond the limit of the collected
data. They are used primarily in science to establish relationships.
- Clearly indicates data correlation (illustrates positive, negative,
strong, weak relationships)
- Method of illustratin non-linear patterns
- Shows spread of data, outliers
- Clearly demonstrate atypical relationships (i.e. spot correlations
are not clear from other graphing methods
- Used for data extrapolation and interpolation
- Too few datapoints can produce skewed interpolations/extrapolations,
producing in correct graph analysis
- Impossible to label data points, hard to ascertain exact values
- Error bars and too many data points can quickly make graph unreadable
- Cannot show relationship between more than two variables at once
A radar graph is composed of a set of axis radiating from a central
origin. Each axis represents a different property or value. Each plot
on the radar graph consists of a point on each of the axis, which are
usually connected. Each plot is assigned a different color. If the
area enclosed by the lines are colored in, the shading is usually
Radar graphs are used when a single plot has more than two properties
(since they allow you to simultaneously display many properties). This
way, plots can be compared over not just two properties. It is easy to
see from a radar graph if a plot is biased towards a particular value,
or is evenly distributed among all the characteristics. It is commonly
used to determine which person an employer should hire. Obviously, the
greater the area covered by the plot, the greater the overall value.
They are used in science, game reviews and employment offices.
- Primary way of displaying more then two or three values at once
- Excellent way to get a "feel" for data
- Cannot compare more than two or three different plots at once
- Without colouring, can be difficult to tell which points belong to who
- With coloring, color could cover up similar looking plots
- Too many axis makes it difficult to read
- Less intuitive than other graph types
A bubble graph is used to display sets of three values: one value is
represented by the bubble's location on the x-axis, one by its
location on the y-axis and the third by its value as represented by
its relative size. They are tyically used to displace financial data,
or to assign importance to a data point.
To better illustrate a bubble graph's use, see:
"In the example, each bubble?s location represents a company?s
sales/revenue information and number of products sold. The sizes of
the bubbles indicate the values of the third variable: the company?s
market share measured in percentages. The chart in this example shows
that company G has the most products and the greatest market share.
However, it does not have the highest sales. "
- Display three variables without using 3D graphs
- Visual size makes it very easy to make relative comparisons
- Conveys same information as a line graph (plus more)
- Due to circle sizes, can be difficult to ascertain actual values (at
best, circle sizes represent estimates)
- Improper scaling can easily skew graph
- Circles may clutter graph, difficult to read and understand
- Cannot be used to display alot of data
Stock graphs are usually a hybrid of various graph types. However, one
type of graph is specific to stocks (seen in example). Each bar
contains two extra dashes, one on the left and on right. Stock graphs
are always seen on a two axis system and are always plotted against
time (time is on the x-axis) and usually a dollar value is used for
The ticks to the left and right serve to indicate the starting and
ending value of the time interval taken up by the bar. The actual
endpoints of the bar are the high and low values reached during that
time period. These special bar graphs are usually combined with
conventional bar graphs or line graphs which show an average value for
each time interval. Stock graphs are used almost exclusively to
visualize financial data.
- Tailored for providing the most pertinent information to investors/traders
- Excellent at displaying financial data
- Sparse labelling for clarification is possible
- Very specific use
- Too many types of graphs combined together can make it very difficult to read
Similar to stock graphs, somtimes graph types are combined together,
such as this pie-bar graph. Depending on the situation, these hybrid
graphs can greatly increase understand or drastically reduce it. The
Could probably have been better represented using just a simple pie
chart. Usually, hybrid graphs can (and should be avoided).
I hope this helps.