How to count unique items in pandas

pandas provides the useful function value_counts() to count unique items – it returns a Series with the counts of unique values.

Category data value count

First of all, let’s see a simple example without any parameter:

import pandas as pd
import numpy as np
# create a dataframe with one column
df = pd.DataFrame({"col1": ["a", "b", "a", "c", "a", "a", "a", "c"]})
# print the dataframe object
print(df)
# line break
print("=" * 30)
# counting unique items
item_counts = df["col1"].value_counts()
print(item_counts)
  • From the output of line 10 you can see the result, which is a count of the column col1.

Category data value count with normalize

Sometimes, we don’t care about the exact number for each item of one column, but we care about the relative percentage. Setting normalize=True can return the percentage instead of the counts.

import pandas as pd
# create a dataframe with one column
df = pd.DataFrame({"col1": ["a", "b", "a", "c", "a", "a", "a", "c"]})
# setting normalize=True
item_counts = df["col1"].value_counts(normalize=True)
print(item_counts)

  • line 7 sets normalize=True.

  • From the output of line 8, you can see the difference from the last demo: the output is a relative percentage, not counts.

Continuous data bucket intervals

value_counts() can handle continuous values as well as category values. By setting bin=n parameters, you can group those continuous values into n groups.

import pandas as pd
import numpy as np
# create a array with random value between 0 and 1
data = np.random.random((30,))
# create a DataFrame object from array
df = pd.DataFrame(data, columns=["col1"])
# show the first five rows of this dataframe object
print(df.head())
# line break
print("=" * 30)
# set bins=8
value_bins = df['col1'].value_counts(bins=8)
print(value_bins)

  • line 12 sets bins=8.

  • From the output of line 13, you can see that the original values are grouped into 8 bins.

Free Resources

License: Creative Commons-Attribution-ShareAlike 4.0 (CC-BY-SA 4.0)