paramcheckup 1.0.3

Check whether the parameters received are correct

paramcheckup

This package has a collection of functions that check whether the parameter received has a certain type, returning True if the input is as expected or raising an error that indicates what the problem is.

Install

pip install paramcheckup

Example 1

import numpy as np
from scipy import stats

Assume a function t_test() that applies one sample Student's t test to compare means (two sided). This function receives three parameters, which are x_data, mu and alpha.

def t_test(x_data, mu, alpha):
    tcalc = (x_data.mean() - mu)*np.sqrt(x_data.size)/(x_data.std(ddof=1))
    t_critical = stats.t.ppf(1-alpha/2, x_data.size - 1)
    p_value = (1 - stats.t.cdf(np.abs(tcalc), x_data.size - 1))*2
    if p_value < alpha:
        conclusion = "Reject H0"
    else:
        conclusion = "Fail to reject H0"
    return tcalc, t_critical, p_value, conclusion

The t_test function strongly depends on the x_data parameter being a one-dimensional NumpyArray. The types.is_numpy(value, param_name, func_name) function can checks whether x_data is in fact a NumpyArray (True):

from paramcheckup import types
def t_test(x_data, mu, alpha):
    types.is_numpy(
        value=x_data,
        param_name="x_data",
        kind="function",
        kind_name="t_test",
        stacklevel=4,
        error=True,
    )
    
    tcalc = (x_data.mean() - mu)*np.sqrt(x_data.size)/(x_data.std(ddof=1))
    t_critical = stats.t.ppf(1-alpha/2, x_data.size - 1)
    p_value = (1 - stats.t.cdf(np.abs(tcalc), x_data.size - 1))*2
    if p_value < alpha:
        conclusion = "Reject H0"
    else:
        conclusion = "Fail to reject H0"
    return tcalc, t_critical, p_value, conclusion

If the user passes a NumpyArray as input for x_data, the result of types.is_numpy will be True and the calculation will be performed as expected:

x = np.array([1.24, 1.3, 1.11])
result = t_test(x, 3, 0.05)
print(result)
(-31.80244895786038, 4.302652729911275, 0.0009872686643235262, 'Reject H0')

However, if you use a list instead of NumpyArray, an TypeError will be raised indicating what the error is:

x = [1.24, 1.3, 1.11]
result = t_test(x, 3, 0.05)
The parameter 'x_data' in function 't_test' must be of type *numpy.ndarray*, but its type is *list*.
UserWarning at line 28: The parameter `x_data` in function `t_test` must be of type `numpy.ndarray`, but its type is `list`.

The UserWarning informs the line where the error occurred, which parameter is wrong and how this parameter should be to be correct. By default, the error traceback is also showed to the user:

Traceback (most recent call last):
  File "...\main.py", line 21, in <module>
    result = t_test(x, 3, 0.05)
  File "...\main.py", line 8, in t_test
    types.is_numpy(x_data, "x_data", "t_test")
  File "...\venv\lib\site-packages\paramcheckup\types.py", line 436, in is_numpy
    raise TypeError("NotNumPyError")
TypeError: NotNumPyError

However, it is possible to silence the traceback through the error parameter:

types.is_numpy(
    value=x_data,
    param_name="x_data",
    kind="function",
    kind_name="t_test",
    stacklevel=4,
    error=False, # <------
)

Note that the function also requires the array to have a single dimension. This could be checked using the paramcheckup.numpy_arrays.n_dimensions function().

Example 2

The alpha parameter indicates the level of significance that should be adopted for the test. It is a value that varies between 0 and 1. To limit the range of values, you can use the numbers.is_between_a_and_b() function:

from paramcheckup import types, numbers

def t_test(x_data, mu, alpha):
    types.is_numpy(
        value=x_data,
        param_name="x_data",
        kind="function",
        kind_name="t_test",
        stacklevel=4,
        error=False,
    )

    numbers.is_between_a_and_b(
        number=alpha,
        lower=0,
        upper=1,
        param_name="alpha",
        kind="function",
        kind_name="t_test",
        inclusive=False,
        stacklevel=4,
        error=False,
    )

    tcalc = (x_data.mean() - mu)*np.sqrt(x_data.size)/(x_data.std(ddof=1))
    t_critical = stats.t.ppf(1-alpha/2, x_data.size - 1)
    p_value = (1 - stats.t.cdf(np.abs(tcalc), x_data.size - 1))*2
    if p_value < alpha:
        conclusion = "Reject H0"
    else:
        conclusion = "Fail to reject H0"
    return tcalc, t_critical, p_value, conclusion


x = np.array([1.24, 1.3, 1.11])
alpha = 1.05
result = t_test(x, 3, alpha)
UserWarning at line 39: The value of `alpha` in function `t_test` must be within the range of `0 < alpha < 1`, but it is `1.05`.

Note that the inclusive=False parameter causes the limits to be open, which makes sense for the significance level. If inclusive=True, we would have obtained the following error:

UserWarning at line 39: The value of `alpha` in function `t_test` must be within the range of `0 <= alpha <= 1`, but it is `1.05`.

Note that alpha must be of numeric type. This could be checked using function paramcheckup.numbers.is_float_or_int() before checking whether the parameter is within a numerical range.

License

• BSD 3-Clause License