The `Lines` Mark

Lines is a Mark object that is primarily used to visualize quantitative data. It works particularly well for continuous data, or when the shape of the data needs to be extracted.

Introduction

The Lines object provides the following features:

Ability to plot a single set or multiple sets of y-values as a function of a set or multiple sets of x-values
Ability to style the line object in different ways, by setting different attributes such as the colors, line_style, stroke_width etc.
Ability to specify a marker at each point passed to the line. The marker can be a shape which is at the data points between which the line is interpolated and can be set through the markers attribute

The Lines object has the following attributes

Attribute	Description	Default Value
`colors`	Sets the color of each line, takes as input a list of any RGB, HEX, or HTML color name	`CATEGORY10`
`opacities`	Controls the opacity of each line, takes as input a real number between 0 and 1	`1.0`
`stroke_width`	Real number which sets the width of all paths	`2.0`
`line_style`	Specifies whether a line is solid, dashed, dotted or both dashed and dotted	`'solid'`
`interpolation`	Sets the type of interpolation between two points	`'linear'`
`marker`	Specifies the shape of the marker inserted at each data point	`None`
`marker_size`	Controls the size of the marker, takes as input a non-negative integer	`64`
`close_path`	Controls whether to close the paths or not	`False`
`fill`	Specifies in which way the paths are filled. Can be set to one of `{'none', 'bottom', 'top', 'inside'}`	`None`
`fill_colors`	`List` that specifies the `fill` colors of each path	`[]`
Data Attribute	Description	Default Value
`x`	abscissas of the data points	`array([])`
`y`	ordinates of the data points	`array([])`
`color`	Data according to which the `Lines` will be colored. Setting it to `None` defaults the choice of colors to the `colors` attribute	`None`

To explore more features, run the following lines of code:

from bqplot import Lines
?Lines

or visit the Lines documentation page

Let's explore these features one by one

We begin by importing the modules that we will need in this example



In [ ]:

    
import numpy as np #For numerical programming and multi-dimensional arrays
from pandas import date_range #For date-rate generation
from bqplot import * #We import the relevant modules from bqplot

Random Data Generation



In [ ]:

    
security_1 = np.cumsum(np.random.randn(150)) + 100.
security_2 = np.cumsum(np.random.randn(150)) + 100.

Basic Line Chart

Using the bqplot, object oriented API, we can generate a Line Chart with the following code snippet:



In [ ]:

    
sc_x = LinearScale()
sc_y = LinearScale()

line = Lines(x=np.arange(len(security_1)), y=security_1,
             scales={'x': sc_x, 'y': sc_y})
ax_x = Axis(scale=sc_x, label='Index')
ax_y = Axis(scale=sc_y, orientation='vertical', label='y-values of Security 1')

Figure(marks=[line], axes=[ax_x, ax_y], title='Security 1')

The x attribute refers to the data represented horizontally, while the y attribute refers the data represented vertically.

We can explore the different attributes by changing each of them for the plot above:



In [ ]:

    
line.colors = ['DarkOrange']

In a similar way, we can also change any attribute after the plot has been displayed to change the plot. Run each of the cells below, and try changing the attributes to explore the different features and how they affect the plot.



In [ ]:

    
# The opacity allows us to display the Line while featuring other Marks that may be on the Figure
line.opacities = [.5]



In [ ]:

    
line.stroke_width = 2.5



In [ ]:

    
line.line_style = 'dashed'



In [ ]:

    
line.interpolation = 'basis'

While a Lines plot allows the user to extract the general shape of the data being plotted, there may be a need to visualize discrete data points along with this shape. This is where the markers attribute comes in.



In [ ]:

    
line.marker = 'triangle-down'

The marker attributes accepts the values square, circle, cross, diamond, square, triangle-down, triangle-up, arrow, rectangle, ellipse. Try changing the string above and re-running the cell to see how each marker type looks.

Plotting a Time-Series

The DateScale allows us to plot time series as a Lines plot conveniently with most date formats.



In [ ]:

    
# Here we define the dates we would like to use
dates = date_range(start='01-01-2007', periods=150)



In [ ]:

    
dt_x = DateScale()
sc_y = LinearScale()

time_series = Lines(x=dates, y=security_1, scales={'x': dt_x, 'y': sc_y})
ax_x = Axis(scale=dt_x, label='Date')
ax_y = Axis(scale=sc_y, orientation='vertical', label='Security 1')

Figure(marks=[time_series], axes=[ax_x, ax_y], title='A Time Series Plot')

Plotting Multiples Sets of Data with `Lines`

The Lines mark allows the user to plot multiple y-values for a single x-value. This can be done by passing an ndarray or a list of the different y-values as the y-attribute of the Lines as shown below.



In [ ]:

    
x_dt = DateScale()
y_sc = LinearScale()

We pass each data set as an element of a list. The colors attribute allows us to pass a specific color for each line.



In [ ]:

    
dates_new = date_range(start='06-01-2007', periods=150)



In [ ]:

    
securities = np.cumsum(np.random.randn(150, 10), axis=0)
positions = np.random.randint(0, 2, size=10)



In [ ]:

    
# We pass the color scale and the color data to the lines
line = Lines(x=dates, y=[security_1, security_2], 
             scales={'x': x_dt, 'y': y_sc},
             labels=['Security 1', 'Security 2'])



In [ ]:

    
ax_x = Axis(scale=x_dt, label='Date')
ax_y = Axis(scale=y_sc, orientation='vertical', label='Security 1')

Figure(marks=[line], axes=[ax_x, ax_y], legend_location='top-left')

Similarly, we can also pass multiple x-values for multiple sets of y-values



In [ ]:

    
line.x, line.y = [dates, dates_new], [security_1, security_2]

Coloring Lines according to data

The color attribute of a Lines mark can also be used to encode one more dimension of data. Suppose we have a portfolio of securities and we would like to color them based on whether we have bought or sold them. We can use the color attribute to encode this information.



In [ ]:

    
x_dt = DateScale()
y_sc = LinearScale()
col_sc = ColorScale(colors=['Red', 'Green'])



In [ ]:

    
dates_color = date_range(start='06-01-2007', periods=150)



In [ ]:

    
securities = 100. + np.cumsum(np.random.randn(150, 10), axis=0)
positions = np.random.randint(0, 2, size=10)
# Here we generate 10 random price series and 10 random positions



In [ ]:

    
# We pass the color scale and the color data to the lines
line = Lines(x=dates_color, y=securities.T, 
             scales={'x': x_dt, 'y': y_sc, 'color': col_sc}, color=positions,
             labels=['Security 1', 'Security 2'])



In [ ]:

    
ax_x = Axis(scale=x_dt, label='Date')
ax_y = Axis(scale=y_sc, orientation='vertical', label='Security 1')

Figure(marks=[line], axes=[ax_x, ax_y], legend_location='top-left')

We can also reset the colors of the Line to their defaults by setting the color attribute to None.



In [ ]:

    
line.color = None

Patches

The fill attribute of the Lines mark allows us to fill a path in different ways, while the fill_colors attribute lets us control the color of the fill



In [ ]:

    
sc_x = LinearScale()
sc_y = LinearScale()

patch = Lines(x=[[0, 2, 1.2], [0.5, 2.5, 1.7], [4,5,6, 6, 5, 4, 3]], 
              y=[[0, 0, 1], [0.5, 0.5, -0.5], [1, 1.1, 1.2, 2.3, 2.2, 2.7, 1.0]],
              fill_colors=['orange', 'blue', 'red'],
              fill='inside',
              stroke_width=10,
              close_path=True,
              scales={'x': sc_x, 'y': sc_y},
              display_legend=True)

Figure(marks=[patch], animation_duration=1000)



In [ ]:

    
patch.fill = 'top'



In [ ]:

    
patch.fill = 'bottom'



In [ ]:

    
patch.opacities = [0.1, 0.2]



In [ ]:

    
patch.x = [[2, 3, 3.2], [0.5, 2.5, 1.7], [4,5,6, 6, 5, 4, 3]]



In [ ]:

    
#patch.fill=['', 'blue']
patch.close_path = False



In [ ]:

The Lines Mark