Useful but Uncommon Pandas Functions: to_period, cumsum, groupby, and Category dtype

In this tutorial we showcase some uncommon yet very handy Pandas functions using a sample DataFrame with three columns (date, class, amount) and 100 rows.

First, we create the DataFrame:

import numpy as np
import pandas as pd
df = pd.DataFrame({
    "date": pd.date_range(start="2021-11-20", periods=100, freq="D"),
    "class": ["A","B","C","D"] * 25,
    "amount": np.random.randint(10, 100, size=100)
})

df.head()

The DataFrame contains a continuous date column, a categorical class column with four distinct values, and a random integer amount column.

1. to_period

The to_period method converts datetime values to a specific time period such as month ("M") or quarter ("Q"), enabling proper time‑series grouping.

df["month"] = df["date"].dt.to_period("M")
df["quarter"] = df["date"].dt.to_period("Q")

df.head()

We can view the counts of each month and quarter:

df["month"].value_counts()
# output
2021-12   31
2022-01   31
2022-02   27
2021-11   11
Freq: M, Name: month, dtype: int64

--------------------------
df["quarter"].value_counts()
# output
2022Q1   58
2021Q4   42
Freq: Q-DEC, Name: quarter, dtype: int64

2. cumsum and groupby

The cumsum function computes the cumulative sum of a column. Applied directly it gives the running total of amount:

df["cumulative_sum"] = df["amount"].cumsum()
df.head()

To obtain cumulative sums per class, we combine groupby with cumsum: df["class_cum_sum"] = df.groupby("class")["amount"].cumsum() Viewing the result for class "A”:

df[df["class"] == "A"].head()

The new column class_cum_sum contains cumulative totals calculated separately for each class.

3. Category dtype

Columns with a limited set of values can be stored as the category dtype, which uses less memory than the default object type.

df.dtypes
# output
date                datetime64[ns]
class                       object
amount                       int64
month                    period[M]
quarter               period[Q-DEC]
cumulative_sum               int64
class_cum_sum                int64

Convert the class column to a categorical type:

df["class_category"] = df["class"].astype("category")
df.dtypes
# output includes
class_category    category

Memory usage comparison shows the categorical column consumes less than half the memory of the object column:

df.memory_usage()
# output
Index               128
date                800
class               800
amount              800
month                800
quarter              800
cumulative_sum      800
class_cum_sum       800
class_category      304
dtype: int64

Although the difference is modest for this small dataset (496 bytes), it scales dramatically with larger data, providing significant space savings.

END