Plot area charts with different colors for time series data based on multiple condtions in Matplotli-CodePudding

Given a small dataset as follows:

         date  value  type
0  2021-07-31   6.32     1
1  2021-08-31   5.93     1
2  2021-09-30   5.43     2
3  2021-10-30   4.72     2
4  2021-11-30   4.23     3
5  2021-12-31   3.96     3

or:

df = pd.DataFrame({'date': {0: '2021-07-31',
  1: '2021-08-31',
  2: '2021-09-30',
  3: '2021-10-30',
  4: '2021-11-30',
  5: '2021-12-31'},
 'value': {0: 6.32, 1: 5.93, 2: 5.43, 3: 4.72, 4: 4.23, 5: 3.96},
 'type': {0: 1, 1: 1, 2: 2, 3: 2, 4: 3, 5: 3}})

I hope to draw area charts based on type, ie., for date where if type=1, 2, 3, I will use color gray, lightpink and skyblue respecitvely.

How could I do that? Thanks.

The expected plot will like this:

Reference code:

year_n_1 = [1.5, 3, 10, 13, 22, 36, 30, 33, 24.5, 15, 6.5, 1.2]
year_n = [2, 7, 14, 17, 20, 27, 30, 38, 25, 18, 6, 1]

plt.fill_between(np.arange(12), year_n_1, color="lightpink",
                 alpha=0.5, label='year N-1')
plt.fill_between(np.arange(12), year_n, color="skyblue",
                 alpha=0.5, label='year N')

plt.legend()
plt.show()

Out:

EDIT:

df = df.set_index('date')
colors = ['gray', 'lightpink', 'skyblue']
plt.fill_between(df['value'], color=colors[type], alpha=0.5)
plt.legend()
plt.show()

Out:

TypeError: list indices must be integers or slices, not type

CodePudding user response：

Try:

# let's use datetime type 
df['date'] = pd.to_datetime(df['date'])

# the colors
colors = ['gray', 'lightpink', 'skyblue']

# loop and plot
fig, ax = plt.subplots()
for i, (t, d) in enumerate(df.groupby('type')):
    d.plot.area(x='date', y='value', label=t, ax=ax, alpha=0.5, color=colors[i])
    
plt.show()

Output: