Useful Series Methods

s.describe() generates a statistical summary in a single call: count, mean, standard deviation, min, 25th percentile (Q1), median (Q2), 75th percentile (Q3), and max. For string Series it returns count, unique, top, and frequency instead. It is the fastest way to get a feel for a column's distribution when first exploring a dataset.

s.value_counts() returns a new Series with the unique values as the index and their occurrence counts as the data, sorted descending by count. Pass normalize=True to get proportions instead of raw counts. This is the first thing to run on a categorical column to understand its distribution.

s.unique() returns a NumPy array of distinct values in their first-occurrence order (unlike value_counts which sorts by frequency). s.nunique() returns the count of distinct values as an integer — the cardinality of the column. Both skip NaN by default. Use nunique(dropna=False) to count NaN as a distinct value.

s.map() applies a function, a dict, or another Series to every element. When you pass a dict, each value is replaced by the corresponding dict value, and values not in the dict become NaN. When you pass a function, it is applied element-wise. map is ideal for recoding categorical labels or applying lookup tables.

s.apply(func) applies a Python callable to each element and collects the results. Unlike map(), it is designed for more complex functions that may return non-scalar objects. For simple element-wise transformations, prefer map() or a vectorized expression; use apply() when the logic is too complex to vectorize directly.

s.sort_values() returns the Series sorted by data values in ascending order (pass ascending=False for descending). s.sort_index() sorts by index labels. Neither modifies the original Series by default. Pass inplace=True to modify in place (less recommended in modern Pandas code that chains operations).

s.isin(values) returns a boolean Series that is True wherever the element is in the provided list or set. This is more readable than multiple | conditions and is significantly faster for large sets. It is commonly used to filter rows belonging to a specific group or to flag records matching a lookup list.

s.between(left, right, inclusive='both') returns a boolean Series that is True where the value falls within the range [left, right]. Both endpoints are inclusive by default. It is more concise than writing (s >= left) & (s <= right) and clearly communicates intent. Useful for filtering numeric ranges like age groups or price bands.

s.head(n) returns the first n elements (default 5) and s.tail(n) returns the last n. These are used constantly when exploring data to peek at the start and end of a long Series without printing thousands of rows. They return a new Series (a slice), so they do not modify the original.

s.rename('new_name') returns a new Series with a different name attribute. To rename the index labels, pass a dict or function: s.rename(index={'old': 'new'}). Renaming is important before combining Series into a DataFrame, because the Series name becomes the column header. Always rename rather than reassign raw values.

s.idxmin() returns the index label of the minimum value and s.idxmax() returns the label of the maximum value. This is more useful than argmin/argmax when the index has meaningful labels like dates or product names — you get the actual identifier, not just a position number.

Test your understanding of useful Series methods from this lesson.

In this lesson you learned: describe() provides a statistical summary in one call, value_counts() builds a frequency table of unique values, and map() translates values using a dict while apply() runs a custom function on each element. Next up we start working with DataFrames — the two-dimensional workhorse of Pandas data analysis.