liveBook · Manning

Type hints

As explained in Software Engineering for Data Scientists

A type hint is an annotation indicating the intended data type of a function parameter or return value without inherently enforcing that type at runtime.

Type hints annotate function parameters and return values with intended data types. They make unfamiliar code easier to read, but they do not enforce types at runtime. section 3.4

Annotating a small function

def add(a: int, b: int) -> int: return a + b The annotations communicate that add expects integers and returns an integer. A separate isinstance check is needed to validate inputs and raise TypeError. section 3.4

Type hints become especially useful when functions are split into data-science pipeline steps. Annotations can identify filenames, pandas data frames and series, numeric thresholds, and predicted outputs. section 3.4.1

Listing 3.31 The code here is a copy of our earlier example in Listing 3.24, except now we add in type hints for each function. This code is also available in the code_modularization_with_type_hints.py file from ch3.
import pandas as pd #1

def read_data(file_name: str) -> pd.DataFrame: #2

    df = pd.read_csv(file_name) #2
    return df #2

def cap_outliers(feature:pd.Series,
        upper_percentile: float, #3
        lower_percentile: float) -> pd.Series: #3

    # get the lower and upper thresholds #3
    # based on the input percentiles #3
    lower_thresh = feature.quantile(lower_percentile) #3
    upper_thresh = feature.quantile(upper_percentile) #3

    # cap any outliers at the lower
    # and upper threshold bounds #3
    feature = feature.\ #3
        map(lambda val: \ #3
        lower_thresh if val < lower_thresh \ #3
        else upper_thresh if val > upper_thresh \ #3
        else val)  #3

    return feature #3


def clean_data(df: pd.DataFrame,
           upper_percentile: float, #4
           lower_percentile: float) -> pd.DataFrame: #4

    df = df.fillna(df.median(numeric_only=True)) #4

    for col in df.columns: #4
        df[col] = cap_outliers(df[col], #4
           upper_percentile, #4
           lower_percentile) #4

    # additional cleaning... #4
    # [code block] #4

    return df #4

if __name__ == "__main__": #5

    customer_data = read_data("customer_data.csv") #5

    cleaned_data = clean_data(customer_data) #5

    # additional code #5
    # [code block] #5

Inspect a modular data-cleaning example annotated with str, pandas DataFrame and Series, and float types. section 3.4.1

How Software Engineering for Data Scientists covers it

How deep it goes
Defines parameter and return annotations, demonstrates them in data functions, and stops short of runtime enforcement.
Examples
Uses add, get_model_predictions, read_data, cap_outliers, and clean_data to demonstrate annotations.
What it assumes
Assumes basic Python functions and data types while teaching annotation syntax and explicit type validation.
What you can do after
Apply type hints to Python functions and add isinstance checks when execution must reject invalid data types.
How it connects
Connects type hints with functions, runtime type checking, pandas DataFrame, pandas Series, and TypeError.

Also explained in

Compare how 22 books explain type hints →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Software Engineering for Data Scientists. The section links point to the parts of the book it draws on.