Data tests

validation_tests.py

def cases_vs_deaths(df):
    """Checks that death count is no more than case count."""
    return (df['deaths'] <= df['cases']).all()


def unique_records(df):
    """Checks that each date and FIPs combination is unique."""
    return df[['date', 'fips']].drop_duplicates().shape[0] == df.shape[0]


def no_nulls_test(df):
    """Checks that all elements are not null"""
    return df.isnull().values.sum() == 0


def range_test(series, min, max):
    """Checks that all values in a series are within a range, inclusive"""
    return (series >= min).all() and (series <= max).all() 


def cases_range_test(df):
    """Checks that all cases are non-negative and <= 10M"""
    return range_test(df['cases'], 0, 10e6)


def deaths_range_test(df):
    """Checks that all deaths are non-negative and <= 100K"""
    return range_test(df['deaths'], 0, 1e5)