Counting unique values in a column in pandas dataframe like in Qlik?

Question

If I have a table like this:

df = pd.DataFrame({
         'hID': [101, 102, 103, 101, 102, 104, 105, 101],
         'dID': [10, 11, 12, 10, 11, 10, 12, 10],
         'uID': ['James', 'Henry', 'Abe', 'James', 'Henry', 'Brian', 'Claude', 'James'],
         'mID': ['A', 'B', 'A', 'B', 'A', 'A', 'A', 'C']
})

I can do count(distinct hID) in Qlik to come up with count of 5 for unique hID. How do I do that in python using a pandas dataframe? Or maybe a numpy array? Similarly, if were to do count(hID) I will get 8 in Qlik. What is the equivalent way to do it in pandas?

Papeete

@piRSquared thanks. I could do something like df[['dID','hID']].agg(['count', 'size', 'nunique']) and it works. But it does not work when combined with groupby. So df[['dID','hID']].groupby('mID').agg(['count', 'size', 'nunique']) says KeyError. is there a way to select particular columns and apply a condition? – Papeete

Three ways df[['mID', 'dID','hID']].groupby('mID').agg(['count', 'size', 'nunique']) – Korfonta

Or df[['dID','hID']].groupby(df['mID']).agg(['count', 'size', 'nunique']) – Korfonta

Or df.groupby('mID')[['dID', 'hID']].agg(['count', 'size', 'nunique']) – Korfonta

Answer 1

Count distinct values, use nunique:

df['hID'].nunique()
5

Count only non-null values, use count:

df['hID'].count()
8

Count total values including null values, use the size attribute:

df['hID'].size
8

Edit to add condition

Use boolean indexing:

df.loc[df['mID']=='A','hID'].agg(['nunique','count','size'])

OR using query:

df.query('mID == "A"')['hID'].agg(['nunique','count','size'])

Output:

nunique    5
count      5
size       5
Name: hID, dtype: int64

Outside

Thanks! How do we add a condition? Like nunique for mID='A'? – Papeete

How to count number of None values? I have a df of only None, and .unique() returns 0 – Santo

@Santo Use. isna like this df['col'].isna().sum() – Outside

How to count distinct values, including nans? meaning count if no nans, or count+1 if any nan exists? – Santo

@Santo To get count of distinct values, you use nunique. There is a parameter dropna default to True, but if you change it to False, it will count the distinct values and add one for NaN. Example, df['val'].nunique(dropna=False). – Outside

Answer 2

If I assume data is the name of your dataframe, you can do :

data['race'].value_counts()

this will show you the distinct element and their number of occurence.

Astromancy

If you want the proportions for each unique item you can also do. data['race'].value_counts(normalize=True) – Plumbing

Answer 3

Or get the number of unique values for each column:

df.nunique()

dID    3
hID    5
mID    3
uID    5
dtype: int64

---

New in pandas 0.20.0 pd.DataFrame.agg

df.agg(['count', 'size', 'nunique'])

         dID  hID  mID  uID
count      8    8    8    8
size       8    8    8    8
nunique    3    5    3    5

---

You've always been able to do an agg within a groupby. I used stack at the end because I like the presentation better.

df.groupby('mID').agg(['count', 'size', 'nunique']).stack()


             dID  hID  uID
mID                       
A   count      5    5    5
    size       5    5    5
    nunique    3    5    5
B   count      2    2    2
    size       2    2    2
    nunique    2    2    2
C   count      1    1    1
    size       1    1    1
    nunique    1    1    1

Korfonta

Thanks! How do we add a condition? Like nunique for mID='A'? – Papeete

@AlhpaDelta I added something at the end. Hope that helps – Korfonta

Answer 4

For unique count of your rows without duplications

df['hID'].nunique()

To know the number of each unique row content duplicated

df['hID'].value_counts()

Hydrolysate

Answer 5

You can use nunique in pandas:

df.hID.nunique()
# 5

Quitrent

Answer 6

To count unique values in column, say hID of dataframe df, use:

len(df.hID.unique())

Isallobar

df.hID.nunique() is much more readable. – Dumb

Answer 7

I was looking for something similar and I found another way you may help you

• If you want to count the number of null values, you could use this function:

def count_nulls(s):
    return s.size - s.count()

• If you want to include NaN values in your unique counts, you need to pass dropna=False to the nunique function.

def unique_nan(s):
    return s.nunique(dropna=False)

• Here is a summary of all the values together using the titanic dataset:

from scipy.stats import mode

agg_func_custom_count = {
    'embark_town': ['count', 'nunique', 'size', unique_nan, count_nulls, set]
}
df.groupby(['deck']).agg(agg_func_custom_count)

You can find more info Here

Wanton

Answer 8

you can use unique property by using len function

len(df['hID'].unique()) 5

Misha