AM

Tutorials Blog Roll of Honour About

 

 

 

Bar, Line and Area Plots with Altair and Python

Standard Imports

import pandas as pd
import altair as alt

from vega_datasets import data

Seattle Weather Data

The Seattle weather dataset is a record of the weather in Seattle, WA, USA, over four years.

df = data.seattle_weather()
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1461 entries, 0 to 1460
Data columns (total 6 columns):
 #   Column         Non-Null Count  Dtype         
---  ------         --------------  -----         
 0   date           1461 non-null   datetime64[ns]
 1   precipitation  1461 non-null   float64       
 2   temp_max       1461 non-null   float64       
 3   temp_min       1461 non-null   float64       
 4   wind           1461 non-null   float64       
 5   weather        1461 non-null   object        
dtypes: datetime64[ns](1), float64(4), object(1)
memory usage: 68.6+ KB

Adding Some Categorical Values

Because date is a datetime type, we can use it to create subcategories. This is useful for the tutorial, but it's also common in business practice, where data is broken down by year, by month, by day, or whatever.

The additional categories are found by calling lambda functions on the date column. There's a full list of the possible date formats here: https://strftime.org/. These formats are common across a lot of languages.

df['year'] = df.date.apply(lambda x: x.strftime('%Y'))
df['month'] = df.date.apply(lambda x: x.strftime('%m'))
df['day'] = df.date.apply(lambda x: x.strftime('%a'))
df['dayOfYear'] = df.date.apply(lambda x: x.strftime('%j'))

df.head()

	date	precipitation	temp_max	temp_min	wind	weather	year	month	day	dayOfYear
0	2012-01-01	0.0	12.8	5.0	4.7	drizzle	2012	01	Sun	001
1	2012-01-02	10.9	10.6	2.8	4.5	rain	2012	01	Mon	002
2	2012-01-03	0.8	11.7	7.2	2.3	rain	2012	01	Tue	003
3	2012-01-04	20.3	12.2	5.6	4.7	rain	2012	01	Wed	004
4	2012-01-05	1.3	8.9	2.8	6.1	rain	2012	01	Thu	005

Bar Charts

Bar charts, as you may remember from school, consist of categories along the horizontal axis and values for those categories along the vertical axis. But how do we group the data by category? If we want to chart precipitation, say, and have year on the x-axis, how do handle all the precipitation data that exists for each year?

Altair Aggregation

The answer is that we have to aggregate the data. There are two ways to do that.

1. Creating a second DataFrame by calling .groupby() on the df DataFame, grouping the data by year and then calling .sum(), count() or whatever on the data, or
2. Using one of Altair's five aggregation methods:
   • Count
   • Sum
   • Min
   • Max
   • Mean

Let's look at the Altair aggregation methods.

1. Count

count() shows the number of days for which we have records - 365 for the years 2013, '14, and '15, and 366 for 2012, a leap year.

c = alt.Chart(df).mark_bar().encode(
    x = 'year',
    y = 'count(precipitation)',
tooltip=['year', 'count(precipitation)'])
c

2. Sum

sum() shows the total precipitation for each year.

c = alt.Chart(df).mark_bar().encode(
    x = 'year',
    y = 'sum(precipitation)',
tooltip=['year', 'sum(precipitation)'])

c

3. Min

min() shows least recorded amount of daily precipitation for each year in the dataset - 0, in each case.

c = alt.Chart(df).mark_bar().encode(
    x = 'year',
    y = 'min(precipitation)',
tooltip=['year', 'min(precipitation)'])

c

4. Max

max() shows the largest recorded amount of daily precipitation for each year in the dataset.

c = alt.Chart(df).mark_bar().encode(
    x = 'year',
    y = 'max(precipitation)',
tooltip=['year', 'max(precipitation)'])

c

5. Mean

mean() shows the average amount of daily precipitation for each year in the dataset.

c = alt.Chart(df).mark_bar().encode(
    x = 'year',
    y = 'mean(precipitation)',
tooltip=['year', 'mean(precipitation)'])

c

Improving the charts

These line charts can be improved in two ways:

1. We can add a color parameter to the .encode() method, thus giving us a stacked chart, and
2. We can swap the x- and y-axes, to turn the chart sideways, and thus more suited to the cells in a notebook.

c = alt.Chart(df).mark_bar().encode(
    y = 'year',
    x = 'sum(precipitation)',
    color='month',
    tooltip=['year', 'month', 'sum(precipitation)'])

c

Line Charts

Line charts are easy. All you need is numerical data for the y-axis and away you go.

c = alt.Chart(df).mark_line().encode(
    x = 'date',
    y = 'precipitation',
tooltip = ['date', 'precipitation']).interactive()
c

Well, maybe not quite. You do need suitable data. 1,461 values are too many values for an x-axis. However, if we reduce the timespan of the data, things get better. We can slice the data to show only the data for January, 2012:

c = alt.Chart(df[(df.date>'2011-12-31')&(df.date<'2012-02-01')]).mark_line().encode(
    x = 'date',
    y = 'precipitation',
tooltip = ['date', 'precipitation']).interactive()
c

Or, we can use the dayOfYear category we cleverly added to our dataset at the start to allow us to compare precipitation in terms of day-on-day, and figure out the exact dates using the tooltip. This is online graph-making at its most useful:

c = alt.Chart(df).mark_line().encode(
    x = 'dayOfYear',
    y = 'precipitation',
    color = 'year',
tooltip = ['date', 'precipitation']).interactive()
c

Line Charts and Aggregation

We can use aggregations in line charts, but they don't always come out as well as we might like.

c = alt.Chart(df).mark_line().encode(
    x = 'month',
    y = 'sum(precipitation)',
    color = 'year',
tooltip = ['date', 'precipitation']).interactive()
c

There is, however, another type of chart in which aggregations work out beautifully, and that gives us the overall "feel" for the data of a bar chart combined with the notion of progress-in-time of a line chart. This is an

Area Chart!

Just substitute .mark_area() for .mark_line() and leave everything else the same:

c = alt.Chart(df).mark_area().encode(
    x = 'month',
    y = 'sum(precipitation)',
    color = 'year',
tooltip = ['month', 'year', 'sum(precipitation)']).interactive()
c

A very beautiful and meaningful result.

The Original Video