The icikt API Reference
Python Information-Content-Informed Kendall Tau Correlation (ICIKT)
The icikt package provides a Python tool to calculate an information-content-informed Kendall Tau correlation coefficient between arrays, while also handling missing values or values which need to be removed.
- icikt.methods.get_global_data(shmName, shape, dtype)[source]
Retrieve global data from shared memory.
- Parameters:
shmName – Name of global data
shape – Shape of global data
dtype – dtype of global data
- Returns:
data ndarray
- icikt.methods.icikt(x: ndarray, y: ndarray, perspective: str = 'global') tuple[source]
Finds missing values, and replaces them with a value slightly smaller than the minimum between both arrays.
- Parameters:
x – First array of data
y – Second array of data
perspective – perspective can be ‘local’ or ‘global’. Default is ‘global’. Global includes (NA,NA) pairs in the calculation, while local does not.
- Returns:
tuple with correlation, pvalue, and tauMax values
- icikt.methods.iciktArray(dataArray: ndarray, globalNA: List[float] = [nan, inf, 0.0], perspective: str = 'global', scaleMax: bool = True, diagGood: bool = True, chunkSize: int = 1, includeOnly: tuple = None) tuple[source]
Calls iciKT to calculate ICI-Kendall-Tau between every combination of columns in the input 2d array, dataArray. Also replaces any instance of the globalNA in the array with np.nan.
- Parameters:
dataArray – 2d array with columns of data to analyze
globalNA – Optional list of values to be considered “missing”. Default is NaN, Inf, and 0.
perspective – perspective can be ‘local’ or ‘global’. Default is ‘global’. Global includes (NA,NA) pairs in the calculation, while local does not.
scaleMax – should everything be scaled compared to the maximum correlation?
diagGood – should the diagonal entries reflect how many entries in the sample were “good”?
chunkSize – What should the size of the chunks be for multiprocessing? Default is 1.
includeOnly – only run correlations of specified columns/combinations
- Returns:
tuple of the output correlations, raw correlations, pvalues, and max tau 2d arrays
Future Parameters: featureNA sampleNA
- icikt.methods.icikt_mp_wrapper(pairwiseIndices: ndarray, perspective: str, shm: SharedMemory, shape: tuple, dtype: dtype) tuple[source]
Wrapper function which is given to multiprocessing. This then calls the icikt method using the indices of pairwise combinations and the perspective.
- Parameters:
pairwiseIndices – Indices of pairwise combination
perspective – perspective can be ‘local’ or ‘global’. Default is ‘global’. Global includes (NA,NA) pairs in the calculation, while local does not.
dtype – dtype in globalShm array
shape – shape of global array
shm – shared memory of global data
- Returns:
tuple result of the icikt method