Skip to content
Snippets Groups Projects
Select Git revision
  • 3f91d376ec4c465eae1d7d5bd4c8a81f653e3a0f
  • master default protected
2 results

univariate.py

Blame
    • univariate.py 2.80 KiB
    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    30
    31
    32
    33
    34
    35
    36
    37
    38
    39
    40
    41
    42
    43
    44
    45
    46
    47
    48
    49
    50
    51
    52
    53
    54
    55
    56
    57
    58
    59
    60
    61
    62
    63
    64
    65
    66
    67
    68
    69
    70
    71
    72
    73
    74
    75
    76
    77 
    

    

    

    ### Univariate tests ### 

import numpy as np
import pandas as pd

# from sklearn.preprocessing import scale
from tools.preprocessing_tools import scale

from tqdm import tqdm

import tools.preprocessing_tools as pt
from tools.compute_manova import linear_regression, ttest


from plotly.subplots import make_subplots
import plotly.graph_objects as go

class UNIVARIATE :
    def __init__(self, outputs, predictors, covariates=None, cols_outputs = None,
                 cols_predictors = None, cols_covariates = None, L_preproc = [], use_resid = True):
        ### Initializing 
        self.outputs = outputs
        self.cols_outputs = cols_outputs
        self.outputs, self.cols_outputs = pt._extract_cols(self.outputs, self.cols_outputs)
        if len(L_preproc)>0:
            self.outputs = pt.pipeline(self.outputs, L_pipe = L_preproc)
        self.outputs = scale(self.outputs)

        self.predictors = predictors
        self.cols_predictors = cols_predictors
        self.predictors, self.cols_predictors = pt._extract_cols(self.predictors, self.cols_predictors)
        self.predictors = scale(self.predictors)



        self.covariates = covariates
        self.cols_covariates = cols_covariates
        if covariates is not None :
            self.covariates, self.cols_covariates = pt._extract_cols(self.covariates, self.cols_covariates) 
            self.covariates = scale(self.covariates) 

        self.use_resid = use_resid
        if use_resid and isinstance(covariates, np.ndarray):
            print("computing residuals")
            self.outputs = np.apply_along_axis(lambda x : pt.adjust_covariates(x,covariates), axis = 0, arr = self.outputs) 

        self.p = None
        self.beta = None

    def test(self, **kwargs):
        nan_mask = np.isnan(self.predictors).flatten()
        if ((self.covariates is not None) and (self.use_resid == False)) :
            print(self.outputs.shape)
            beta = linear_regression(self.outputs[~nan_mask], self.predictors[~nan_mask], self.covariates[~nan_mask])
        else :
            beta = linear_regression(self.outputs[~nan_mask], self.predictors[~nan_mask])
        self.beta = beta
        res = ttest(beta, self.predictors[~nan_mask], self.outputs[~nan_mask])
        self.p = res

    def plot(self, show = True):
        fig = make_subplots(rows = 1,cols=1)
        fig.update_layout(title = "Plot of all univariate tests")

        fig.add_trace(
            go.Scatter(x = self.cols_outputs, y=-np.log10(self.p), name = "Univariate tests", mode='markers'),
            row=1,col=1
        )
        fig.add_trace(
            go.Scatter(x = self.cols_outputs, y = [-np.log10(0.05/len(self.cols_outputs))]*len(self.cols_outputs), name = 'Bonferroni threshold'),
            row=1,col=1
        )

        if show:
            fig.show()
        else :
            return fig