pandas overlay on top of a plot

Question

I have some time series data in a pandas dataframe that I can visualise as follows:

import pandas as pd

d = {'end_time': [datetime.datetime(2020, 3, 17, 0, 0), datetime.datetime(2020, 3, 17, 0, 5), datetime.datetime(2020, 3, 17, 0, 10), datetime.datetime(2020, 3, 17, 0, 15), datetime.datetime(2020, 3, 17, 0, 20), datetime.datetime(2020, 3, 17, 0, 25), datetime.datetime(2020, 3, 17, 0, 30), datetime.datetime(2020, 3, 17, 0, 35), datetime.datetime(2020, 3, 17, 0, 40), datetime.datetime(2020, 3, 17, 0, 45), datetime.datetime(2020, 3, 17, 0, 50), datetime.datetime(2020, 3, 17, 0, 55)], "measurement": [2000, 1500, 800, 900, 400, 4000, 300, 900, 1000, 1250, 1100, 1300], "reliability": [99, 81, 84, 85, 99, 86, 96, 97, 98, 99, 98, 97]}

# select some relevant columns
subset_df = pd.DataFrame.from_dict(d)

# plot measurements over time
subset_df.plot('end_time', 'measurement')

Now the reliabilitycolumn is a number between 0 and 100. What I want to do is highlight areas where this reliability score is below 95. So something where I can overlay a transparent box around those areas to visually highlight where the measurements may not be very reliable.

Asked By: Luca

||

Source

Answer 1

Considering this random dataframe, where

end_time: dates starting from 2020-03-17 00:00:00 to 2020-03-17 00:55:00 with a 5 minutes interval
measurement: random integers between 300 and 4000

reliability: random integers between 0 and 100

import pandas as pd
import numpy as np

df = pd.DataFrame({'end_time': pd.date_range(start='2020-03-17 00:00:00', end='2020-03-17 00:55:00', freq='5min'),
                   'measurement': np.random.randint(300, 4000, size=12),
                      'reliability': np.random.randint(0, 100, size=12)})

[Out]:

              end_time  measurement  reliability
0  2020-03-17 00:00:00         3905            7
1  2020-03-17 00:05:00         1143           93
2  2020-03-17 00:10:00         2672           55
3  2020-03-17 00:15:00          416           29
4  2020-03-17 00:20:00         1246           21
5  2020-03-17 00:25:00         2743           32
6  2020-03-17 00:30:00         2798           49
7  2020-03-17 00:35:00         1012           21
8  2020-03-17 00:40:00         3894           64
9  2020-03-17 00:45:00         1218           18
10 2020-03-17 00:50:00         1600           97
11 2020-03-17 00:55:00          729           76

If the goal is to plot all measures with a reliability lower than 95 in red and the rest in blue, let us first create a few variables that will be useful:

measurement with a reliability lower than 95:

measures = df[df.reliability < 95].measurement

end_time of the measurement with a reliability lower than 95:
```
dates = df[df.reliability < 95].end_time
```

measurement with a reliability higher than 95:

measures2 = df[df.reliability >= 95].measurement

end_time of the measurement with a reliability higher than 95:
```
dates2 = df[df.reliability >= 95].end_time
```

Now let us create the plot

import matplotlib.pyplot as plt

# Create the plot:
plt.plot(dates, measures, 'ro', dates2, measures2, 'bo')
# Set the title:
plt.title('Measures over time')
# Set the x label:
plt.xlabel('Date')
# Set the y label:
plt.ylabel('Measure')
# Set the x ticks:
plt.xticks(rotation=45)
# Show the plot:
plt.show()

Now, as per requirement (use fill_between so that I can paint a transparent box from the x-axes to the top of y-axes), before plt.show() one can use the following

plt.fill_between(dates, 0, measures, color='red', alpha=0.2)

Answered By: Gonçalo Peres

pandas overlay on top of a plot

Question:

Answers: