Seaborn heatmap between date column and integer column

Question

I have a Dataframe containing 'Date' column and 'Tweet_Count' column.

I want to plot a heatmap that will show tweet counts on each day of one month.

Using this code:

uk = uk_df.pivot("Date", "Tweet_Count")
ax = sns.heatmap(uk)

I am getting an error:

TypeError: '<=' not supported between instances of 'float' and 'str'

Any help would be really appreciated. Thank you.

user_6366770 · Answer

The required DataFrame format for a sns.heatmapyou are looking for is a pivot table with an index, columns and values passed as Date, Country and Tweet_Count to pivot, respectively, below (if this causes an error, please upgrade to the latest version of pandas as pivot had issues in past versions):

   Country    UK
      Date  
2020-03-01  400
2020-03-02  1000
2020-03-03  100

So, you simply need to pass Country to pivot as well:

uk = pd.DataFrame({'Country': {0: 'UK', 1: 'UK', 2: 'UK'},
 'Date': {0: '2020-03-01', 1: '2020-03-02', 2: '2020-03-03'},
 'Tweet_Count': {0: 100, 1: 200, 2: 300}})

uk['Date'] = pd.to_datetime(uk['Date']).dt.date

uk = uk.pivot("Date", "Country", 'Tweet_Count')
ax = sns.heatmap(uk)

Scroll down to the fourth block of code here:

It's worth mentioning that you can also annotate with:

ax = sns.heatmap(uk, annot=True, fmt="d")

user_12258459 · Answer

Let's say the data looks like this, below I have two countries

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

Date = pd.date_range(start='1/3/2020', periods=30,freq="D").strftime('%Y-%m-%d').to_list()
df = pd.DataFrame({'Country':np.repeat(['UK','KU'],len(Date)),
                   'Date':Date*2,
                     'Tweet_Count':np.random.randint(1000,2000,len(Date)*2)})
df.head()

    Country Date    Tweet_Count
0   UK  2020-01-03  1809
1   UK  2020-01-04  1419
2   UK  2020-01-05  1463
3   UK  2020-01-06  1576
4   UK  2020-01-07  1137

If there's only 1 country, then a transpose will work:

fig, ax = plt.subplots(figsize=(8,2))
uk = df[df['Country']=="UK"]
ax = sns.heatmap(uk[['Date','Tweet_Count']].set_index('Date').T)

If there's 2 or more:

wide_df = df.pivot_table(index="Country",columns="Date",fill_value="Tweet_Count")
wide_df.columns = [j for i,j in wide_df.columns]
fig, ax = plt.subplots(figsize=(8,2))
ax = sns.heatmap(wide_df)

Seaborn heatmap between date column and integer column

2 Answers