CODEWITHSUNDEEP
pythonmatplotlibdata-sciencevisualization
oskar szarowicz
oskar szarowicz

Reputation: 133

How to helpfully plot time series data in python

I have a 2 Dimensional set of time series data containing about 1000 samples. I.e I have a long list of 1000 elements each entry is a list of two numbers.

It can be thought of the position, x and y coordinates, of a car with a picture taken each second for 1000 seconds. When I plot this, as seen below, you get a decent idea of the trajectory but it's unclear where the car starts or finishes, i.e which direction it is traveling in. I was thinking about including arrows between each point but I think this would get quite clustered (maybe you know a way to overcome that issue?) Also, I thought of colouring each point with a spectrum that made it clear to see time increasing, i.e hotter points to colder points as time goes on. Any idea how to achieve this in matplotlib? enter image description here

Upvotes: 0

Views: 107

Answers (1)

fr_slott
fr_slott

Reputation: 131

I believe both your ideas would work well, I just think you need to test which option works best for your case.

Option 1: arrows

To avoid a cluttered plot I believe you could plot arrows between only a selection of points to show the general direction of your trajectory. In my example below I only plot an arrow between points 1 and 2, 6 and 7, and so on and. You might want to increase the spacing between the points to make this work for your long series. It is also possible to connect points that are seperated by, say, 10 points to make them more clearly visible.

import numpy as np
import matplotlib.pyplot as plt

# example data
x = np.linspace(0, 10, 100)
y = x

plt.figure()

# plot the data points
for i in range(len(x)):
    plt.plot(x[i], y[i], "ro")

# plot arrows between points 1 and 2, 6 and 7 and so on.
for i in range(0, len(x)-1, 5):
    plt.arrow(x[i], y[i], x[i+1] - x[i], y[i+1] - y[i], color = "black",zorder = 2, width = 0.05)

plt.xlabel('x')
plt.ylabel('y')
plt.show()

This yields this plot.

Option 2: colors

You can generate any number of colors from a colormap, meaning you can make a list of 1000 sequential colors. This way you can plot each of your points in an increasingly warm color.

Example:

import numpy as np
import matplotlib.pyplot as plt

# example data
x = np.linspace(0, 10, 100)
y = x

# generate 100 (number of data points) colors from colormap
colors = [plt.get_cmap("coolwarm")(i) for i in np.linspace(0,1, len(x))]

plt.figure()

# plot the data points with the generated colors
for i in range(len(x)):
    plt.plot(x[i], y[i], color = colors[i], marker = "o")

plt.xlabel('x')
plt.ylabel('y')
plt.show()

This yields this figure, where the oldest data point is cool (blue) and the newest is red (warm).

Upvotes: 2

Related Questions