calfpy 1.20.1

Contains greedy algorithms for coarse approximation linear functions.

CALF v1.20.1 README

Setup and Usage

You can install CALF by running pip install calfpy

You can import all CALF methods by using the folliwng import statement
from calfpy.methods import *

Calling a function (example) calf(data, 3, "binary", optimize = 'pval', verbose = False)

Library Documentation

calf(data, nMarkers, targetVector, optimize = 'pval', verbose = False)

Coarse Approximation Linear function. The main function used to invoke the CALF algorithm on a dataset.

parameter	description	type
data	Data frame where first column contains case/control coded variable (0/1) if targetVector is binary or real values if targetVector is nonbinary	pandas DataFrame
nMarkers	Maximum number of markers to include in creation of sum.	int
targetVector	Set to "binary" to indicate data with a target vector (first column) having case/control characteristic. Set to "nonbinary" for target vector (first column) with real numbers.	string
optimize	Criteria to optimize. Allowed values are "pval", "auc", for a binary target vector, or "corr" for a nonbinary target vector.	string
verbose	True to print activity at each iteration to console. Defaults to False.	bool

returns

A dictionary composed of the following results from CALF:

key	description/value
selection	The markers selected with their assigned weight (-1 or 1).
auc	The AUC determined during running CALF. AUC can be provided for given markers AUC represented for selected markers will only be optimal if set to optimzie for AUC.
randomize	False
proportion	Undefined
targetVec	Target vector argument given in the function call, 'rocPlot': Receiver operating curve plot, if applicable for dataset type and optimizer supplied.
finalBest	The optimal value for the provided optimization type, e.g. if optimize='pval" this will have the calculated p-value for the run.
optimize	The optimizer argument given in the function call.

calf_fractional(data, nMarkers, controlProportion = .8, caseProportion = .8, optimize = "pval", verbose = False)

Randomly selects from binary input provided to data parameter while ensuring the requested proportions of case and control variables are used and runs Coarse Approximation Linear Function

parameter	description	type
data	Data frame where first column contains case/control coded variables (0/1) (binary data).	pandas DataFrame
nMarkers	Maximum number of markers to include in creation of sum.	int
controlProportion	Proportion of control samples to use, default is .8.	float
caseProportion	Proportion of case samples to use, default is .8.	float
optimize	Criteria to optimize. Allowed values are "pval" or "auc".	string
verbose	True to print activity at each iteration to console. Defaults to False.	bool

returns

A dictionary composed of the following results from CALF:

key	description/value
selection	The markers selected each with their assigned weight (-1 or 1).
auc	The AUC determined during running CALF. AUC can be provided for given markers AUC represented for selected markers will only be optimal if set to optimzie for AUC.
randomize	False
proportion	The proportions of case an control applied druing the function run.
targetVec	"binary"
rocPlot	Receiver operating curve plot, if applicable for dataset type and optimizer supplied.
finalBest	The optimal value for the provided optimization type, e.g. if optimize='pval" this will have the calculated p-value for the run.
optimize	The optimizer argument given in the function call.

calf_randomize(data, nMarkers, targetVector, times = 1, optimize = "pval", verbose = False)

Randomly selects from input provided to data parameter and runs Coarse Approximation Linear Function

parameter	description	type
data	Data frame where first column contains case/control coded variable (0/1) if targetVector is binary or real values if targetVector is nonbinary	pandas DataFrame
nMarkers	Maximum number of markers to include in creation of sum.	int
times	Indicates the number of replications to run with randomization.	int
optimize	Criteria to optimize. Allowed values are "pval", "auc", for a binary target vector, or "corr" for a nonbinary target vector.	string
verbose	True to print activity at each iteration to console. Defaults to False.	bool

returns

A dictionary composed of the following results from CALF:

key	description/value
multiple	The markers chosen and the number of times they were selected per iteration,
auc	The AUC determined during running CALF. AUC can be provided for given markers AUC represented for selected markers will only be optimal if set to optimzie for AUC,
randomize	True
targetVec	"binary"
aucHist	A historgram of the AUC values calcuted for all the iterations,
times	The value provided to the times parameter when the function was called,
rocPlot	Receiver operating curve plot, if applicable for dataset type and optimizer supplied,
finalBest	The optimal value for the provided optimization type, e.g. if optimize='pval" this will have the calculated p-value for the run,
optimize	The optimizer argument given in the function call,
verbose	The value supplied to the verbose parameter when the function was called

calf_subset (data, nMarkers, targetVector, proportion = .8, times = 1, optimize = "pval", verbose = False)

Randomly selects a subset of the data on which to run Coarse Approximation Linear Function

parameter	description	type
param data	Data frame where first column contains case/control coded variable (0/1) if targetVector is binary or real values if targetVector is nonbinary	pandas DataFrame
param nMarkers	Maximum number of markers to include in creation of sum.	int
param targetVector	Set to "binary" to indicate data with a target vector (first column) having case/control characteristic. Set to "nonbinary" for target vector (first column) with real numbers.	string
param proportion	A value between 0 and 1, the percentage of data, randomly chosen, to use in the calculation. Default is .8.	float
param times	Indicates the number of replications to run with randomization.	int
param optimize	Criteria to optimize. Allowed values are "pval", "auc", for a binary target vector, or "corr" for a nonbinary target vector.	string
param verbose	True to print activity at each iteration to console. Defaults to False.	bool

returns

A dictionary composed of the following results from CALF:

key	description/value
multiple	The markers chosen and the number of times they were selected per iteration.
auc	The AUC determined during running CALF. AUC can be provided for given markers AUC represented for selected markers will only be optimal if set to optimzie for AUC.
proportion	The value supplied to the proportion paremeter when calling the function.
targetVec	"binary"
aucHist	A historgram of the AUC values calcuted for all the iterations.
times	The value provided to the times parameter when the function was called.
rocPlot	Receiver operating curve plot, if applicable for dataset type and optimizer supplied.
finalBest	The optimal value for the provided optimization type, e.g. if optimize=pval" this will have the calculated p-value for the run.
optimize	The optimizer argument given in the function call.

calf_exact_binary_subset(data, nMarkers, nCase, nControl, times = 1, optimize = "pval", verbose = False)

Randomly selects subsets of data, case and control, from a binary data set, while precisely ensuring the size of the sets on which to run Coarse Approximation Linear Function

parameter	description	type
data	Data frame where first column contains case/control coded variable (0/1).	pandas DataFrame
nMarkers	Maximum number of markers to include in creation of sum.	int
nCase	The number of data points to use for the set of case samples.	int
nControl	The number of data points to use for the set of control samples.	int
times	Indicates the number of replications to run with randomization	int
optimize	Criteria to optimize. Allowed values are "pval" or "auc"	string
verbose	True to print activity at each iteration to console. Defaults to False.	bool

returns

A dictionary composed of the following results from CALF:

key	description/value
multiple	The markers chosen and the number of times they were selected per iteration.
auc	The AUC determined during running CALF. AUC can be provided for given markers AUC represented for selected markers will only be optimal if set to optimzie for AUC.
proportion	The value supplied to the proportion paremeter when calling the function.
targetVec	"binary"
aucHist	A historgram of the AUC values calcuted for all the iterations.
times	The value provided to the times parameter when the function was called.
rocPlot	Receiver operating curve plot, if applicable for dataset type and optimizer supplied.
finalBest	The optimal value for the provided optimization type, e.g. if optimize='pval" this will have the calculated p-value for the run.
optimize	The optimizer argument given in the function call.

calf_cv(data = CaseControl, limit = 5, times = 100, targetVector = 'binary', optimize = 'pval')

Performs repeated random subsampling cross validation on data for Coarse Approximation Linear Function

parameter	description	type
data	Data frame where first column contains case/control coded variable (0/1) if targetVector is binary or real values if targetVector is nonbinary	pandas DataFrame
limit	Maximum number of markers to attempt to determine per iteration.	int
times	Indicates the number of replications to run with randomization.	int
proportion	A value between 0 and 1, the percentage of data, randomly chosen, to use in each iteration of CALF. Default is .8,	float
optimize	Criteria to optimize. Allowed values are "pval", "auc", for a binary target vector, or "corr" for a nonbinary target vector.	string
outputPath	The path where files are to be written as output, default is None meaning no files will be written. When targetVector is "binary" file binary.csv will be output in the provided path, showing the reults. When targetVector is "nonbinary" file nonbinary.csv will be output in the provided path, showing the results. In the same path, the kept and excluded variables from the LAST iteration, will be output, prefixed with the targetVector type "binary" or "nonbinary" followed by Kept and Excluded and suffixed with .csv. Two files containing the results from each run have List in the filenames and suffixed with .txt.	string

returns

A data frame of the results from the cross validation. Columns of all markers from data and rows representing each iteration of a CALF run. Cells will contain the result from CALF for a given CALF run and the markers that were chose for that run.

perm_target_cv(data, targetVector, limit, times, proportion = .8, optimize = 'pval', outputPath=None)

Performs repeated random subsampling cross validation on data but randomly permutes the target column (first column) with each iteration, for Coarse Approximation Linear Function

parameter	description	type
data	Data frame where first column contains case/control coded variable (0/1) if targetVector is binary or real values if targetVector is nonbinary values if targetVector is nonbinary	pandas DataFrame
limit	Maximum number of markers to attempt to determine per iteration.	int
times	Indicates the number of replications to run with randomization.	int
proportion	A value between 0 and 1, the percentage of data, randomly chosen, to use in each iteration of CALF. Default is .8,	float
optimize	Criteria to optimize. Allowed values are "pval", "auc", for a binary target vector, or "corr" for a nonbinary target vector.	string
outputPath	The path where files are to be written as output, default is None meaning no files will be written. When targetVector is "binary" file binary.csv will be output in the provided path, showing the reults. When targetVector is "nonbinary" file nonbinary.csv will be output in the provided path, showing the results. In the same path, the kept and excluded variables from the LAST iteration, will be output, prefixed with the targetVector type "binary" or "nonbinary" followed by Kept and Excluded and suffixed with .csv. Two files containing the results from each run have List in the filenames and suffixed with .txt.	string

returns

A data frame of the results from the cross validation. Columns of all markers from data and rows representing each iteration of a CALF run. Cells will contain the result from CALF for a given CALF run and the markers that were chose for that run.

write_calf(x, filename)

Writes the results from a call to calf() to a file

parameter	description	type
x	The dictionary object returned from calling calf().	dict
filename	The name of the file in which to write the results from calf().	string

write_calf_randomize(x, filename)

Writes the results from a call to calf_randomize() to a file

parameter	description	type
x	The dictionary object returned from calling calf_randomize().	dict
filename	The name of the file in which to write the results from calf_randomize().	string

write_calf_subset(x, filename)

Writes the results from a call to calf_subset() to a file

parameter	description	type
x	The dictionary object returned from calling calf_subset().	dict
filename	The name of the file in which to write the results from calf_subset().	string

calf_internal(data, nMarkers, randomize = False, proportion = None, times = 1, targetVector = 'binary', optimize = 'pval', verbose = False)

The basic CALF algorithm

parameter	description	type
data	Data frame where first column contains case/control coded variable (0/1) if targetVector is binary or real values if targetVector is nonbinary	pandas DataFrame
nMarkers	Maximum number of markers to include in creation of sum.	int
randomize	Set to True to randomize the data for each CALF run.	bool
proportion	A value between 0 and 1, the percentage of data, randomly chosen, to use in the calculation.	float
times	The number of times to run CALF on data.	int
targetVector	Set to "binary" to indicate data with a target vector (first column) having case/control characteristic. Set to "nonbinary" for target vector (first column) with real numbers.	string
optimize	Criteria to optimize. Allowed values are "pval", "auc", for a binary target vector, or "corr" for a nonbinary target vector.	string
verbose	True to print activity at each iteration to console. Defaults to False.	* bool*

returns

A dictionary composed of the following results from CALF:

key	description/value
selection	The markers selected each with their assigned weight (-1 or 1).
auc	The AUC determined during running CALF. AUC can be provided for given markers AUC represented for selected markers will only be optimal if set to optimzie for AUC.
randomize	False
proportion	Undefined
targetVec	Target vector argument given in the function call.
rocPlot	Receiver operating curve plot, if applicable for dataset type and optimizer supplied.
finalBest	The optimal value for the provided optimization type, e.g. if optimize='pval" this will have the calculated p-value for the run.
optimize	The optimizer argument given in the function call.