Automated_classification_of_stroke_deficit/scripts/figures.py

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from master_funcs import *
from collections import Counter
from imblearn.over_sampling import SMOTE
from imblearn.under_sampling import RandomUnderSampler
from imblearn.pipeline import Pipeline
import dill
#%matplotlib qt
import scipy
import seaborn as sn
import pingouin as pg
import warnings
import dataframe_image as dfi
warnings.filterwarnings("ignore")


## Manual Rater Results Table
# ============================================================================================================================================
mr_cyl_1=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Jan_figure.csv", index_col=0, delimiter=',', header=0)
mr_cyl_2=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Nicole_figure.csv", index_col=0, delimiter=',', header=0)
mr_cyl_3=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Jule_figure.csv", index_col=0, delimiter=',', header=0)

mr_gw_1=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_GridWalk_Analyse_Jan_figure.csv", index_col=0, delimiter=',', header=0)
mr_gw_2=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_GridWalk_Analyse_Nicole_figure.csv", index_col=0, delimiter=',', header=0)
mr_gw_3=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_GridWalk_Analyse_Jule_figure.csv", index_col=0, delimiter=',', header=0)

dfi.export(mr_cyl_1, '/home/user/owncloud/thesis_figures/MR_CY_1.png', max_cols=51)
dfi.export(mr_cyl_2, '/home/user/owncloud/thesis_figures/MR_CY_2.png', max_cols=51)
dfi.export(mr_cyl_3, '/home/user/owncloud/thesis_figures/MR_CY_3.png', max_cols=51)

dfi.export(mr_gw_1, '/home/user/owncloud/thesis_figures/MR_GW_1.png')
dfi.export(mr_gw_2, '/home/user/owncloud/thesis_figures/MR_GW_2.png')
dfi.export(mr_gw_3, '/home/user/owncloud/thesis_figures/MR_GW_3.png')
# ============================================================================================================================================



## Correlation Feature Footfault within MR and between MR and DLC plus FI 2D and 3D
# ============================================================================================================================================
mr_cyl_1=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_GridWalk_Analyse_Jan.csv", index_col=0, delimiter=',', header=None)
mr_cyl_2=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_GridWalk_Analyse_Nicole.csv", index_col=0, delimiter=',', header=None)
mr_cyl_3=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_GridWalk_Analyse_Jule.csv", index_col=0, delimiter=',', header=None)

patients=['PB_T2_3_1', 'PB_T2_3_2', 'PB_T2_3_3', 'PB_T2_4_2', 'PB_T2_5_1', 'PB_T2_6_1', 'PB_T2_6_2', 'PB_T3_23_1', 'PB_T3_23_3', 'PB_T3_24_1', 'PB_T3_24_2', 'PB_T3_24_3']

max_i=np.array([np.nanmax(mr_cyl_1.loc['PB_T2_3_1']), np.nanmax(mr_cyl_2.loc['PB_T2_3_1']), np.nanmax(mr_cyl_3.loc['PB_T2_3_1']), np.nanmax(footfault_2d.loc['PB_T2_3_1']), np.nanmax(footfault_FI_2d.loc['PB_T2_3_1']), np.nanmax(footfault_3d.loc['PB_T2_3_1']), np.nanmax(footfault_FI_3d.loc['PB_T2_3_1'])])
mr_cyl_1_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
mr_cyl_2_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
mr_cyl_3_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
footfault_2d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
footfault_FI_2d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
footfault_3d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
footfault_FI_3d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))

for i in mr_cyl_1.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_1_corr[int(i-15):int(i+16)]=1
for i in mr_cyl_2.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_2_corr[int(i-15):int(i+16)]=1
for i in mr_cyl_3.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_3_corr[int(i-15):int(i+16)]=1
for i in footfault_2d.loc['PB_T2_3_1']:
    if i>0:
        footfault_2d_corr[int(i-15):int(i+16)]=1
for i in footfault_FI_2d.loc['PB_T2_3_1']:
    if i>0:
        footfault_FI_2d_corr[int(i-15):int(i+16)]=1
for i in footfault_3d.loc['PB_T2_3_1']:
    if i>0:
        footfault_3d_corr[int(i-15):int(i+16)]=1
for i in footfault_FI_3d.loc['PB_T2_3_1']:
    if i>0:
        footfault_FI_3d_corr[int(i-15):int(i+16)]=1
    
for z in patients[1:]:
    max_i=np.array([np.nanmax(mr_cyl_1.loc[z]), np.nanmax(mr_cyl_2.loc[z]), np.nanmax(mr_cyl_3.loc[z]), np.nanmax(footfault_2d.loc[z]), np.nanmax(footfault_FI_2d.loc[z]), np.nanmax(footfault_3d.loc[z]), np.nanmax(footfault_FI_3d.loc[z])])
    if all(max_i==max_i):
        mr_cyl_1_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        mr_cyl_2_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        mr_cyl_3_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        footfault_2d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        footfault_FI_2d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        footfault_3d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        footfault_FI_3d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        for i in mr_cyl_1.loc[z]:
            if i>0:
                mr_cyl_1_corr[int(i-15):int(i+16)]=1
        for i in mr_cyl_2.loc[z]:
            if i>0:
                mr_cyl_2_corr[int(i-15):int(i+16)]=1
        for i in mr_cyl_3.loc[z]:
            if i>0:
                mr_cyl_3_corr[int(i-15):int(i+16)]=1
        for i in footfault_2d.loc[z]:
            if i>0:
                footfault_2d_corr[int(i-15):int(i+16)]=1
        for i in footfault_FI_2d.loc[z]:
            if i>0:
                footfault_FI_2d_corr[int(i-15):int(i+16)]=1
        for i in footfault_3d.loc[z]:
            if i>0:
                footfault_3d_corr[int(i-15):int(i+16)]=1
        for i in footfault_FI_3d.loc[z]:
            if i>0:
                footfault_FI_3d_corr[int(i-15):int(i+16)]=1
        mr_cyl_1_corr=np.r_[mr_cyl_1_corr, mr_cyl_1_corr_temp]
        mr_cyl_2_corr=np.r_[mr_cyl_2_corr, mr_cyl_2_corr_temp]
        mr_cyl_3_corr=np.r_[mr_cyl_3_corr, mr_cyl_3_corr_temp]
        footfault_2d_corr=np.r_[footfault_2d_corr, footfault_2d_corr_temp]
        footfault_FI_2d_corr=np.r_[footfault_FI_2d_corr, footfault_FI_2d_corr_temp]
        footfault_3d_corr=np.r_[footfault_3d_corr, footfault_3d_corr_temp]
        footfault_FI_3d_corr=np.r_[footfault_FI_3d_corr, footfault_FI_3d_corr_temp]

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_1_corr, mr_cyl_2_corr), jaccard(mr_cyl_1_corr, mr_cyl_3_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_2_corr), 1, jaccard(mr_cyl_2_corr, mr_cyl_3_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, mr_cyl_3_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(3, 3))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_MR_GW.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_1_corr, mr_cyl_2_corr), jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_1_corr, footfault_2d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_2_corr), 1, jaccard(mr_cyl_2_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, footfault_2d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, mr_cyl_3_corr), 1, jaccard(mr_cyl_3_corr, footfault_2d_corr)],
                       [jaccard(mr_cyl_1_corr, footfault_2d_corr), jaccard(mr_cyl_2_corr, footfault_2d_corr), jaccard(mr_cyl_3_corr, footfault_2d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_GW_2d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_1_corr, mr_cyl_2_corr), jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_1_corr, footfault_FI_2d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_2_corr), 1, jaccard(mr_cyl_2_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, footfault_FI_2d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, mr_cyl_3_corr), 1, jaccard(mr_cyl_3_corr, footfault_FI_2d_corr)],
                       [jaccard(mr_cyl_1_corr, footfault_FI_2d_corr), jaccard(mr_cyl_2_corr, footfault_FI_2d_corr), jaccard(mr_cyl_3_corr, footfault_FI_2d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_GW_FI_2d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_1_corr, mr_cyl_2_corr), jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_1_corr, footfault_3d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_2_corr), 1, jaccard(mr_cyl_2_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, footfault_3d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, mr_cyl_3_corr), 1, jaccard(mr_cyl_3_corr, footfault_3d_corr)],
                       [jaccard(mr_cyl_1_corr, footfault_3d_corr), jaccard(mr_cyl_2_corr, footfault_3d_corr), jaccard(mr_cyl_3_corr, footfault_3d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_GW_3d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_1_corr, mr_cyl_2_corr), jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_1_corr, footfault_FI_3d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_2_corr), 1, jaccard(mr_cyl_2_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, footfault_FI_3d_corr)], 
                       [jaccard(mr_cyl_1_corr, mr_cyl_3_corr), jaccard(mr_cyl_2_corr, mr_cyl_3_corr), 1, jaccard(mr_cyl_3_corr, footfault_FI_3d_corr)],
                       [jaccard(mr_cyl_1_corr, footfault_FI_3d_corr), jaccard(mr_cyl_2_corr, footfault_FI_3d_corr), jaccard(mr_cyl_3_corr, footfault_FI_3d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_GW_FI_3d.png", format='png', dpi=200)



print('GW 3D')
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_1_corr))]+[i for i in range(len(mr_cyl_2_corr))]+[i for i in range(len(mr_cyl_3_corr))], 'rater':['1']*len(mr_cyl_1_corr)+['2']*len(mr_cyl_2_corr)+['3']*len(mr_cyl_3_corr), 'rating':np.concatenate((mr_cyl_1_corr, mr_cyl_2_corr, mr_cyl_3_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_1_corr))]+[i for i in range(len(mr_cyl_2_corr))]+[i for i in range(len(mr_cyl_3_corr))]+[i for i in range(len(footfault_2d_corr))], 'rater':['1']*len(mr_cyl_1_corr)+['2']*len(mr_cyl_2_corr)+['3']*len(mr_cyl_3_corr)+['DLC']*len(footfault_2d_corr), 'rating':np.concatenate((mr_cyl_1_corr, mr_cyl_2_corr, mr_cyl_3_corr, footfault_2d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_1_corr))]+[i for i in range(len(mr_cyl_2_corr))]+[i for i in range(len(mr_cyl_3_corr))]+[i for i in range(len(footfault_FI_2d_corr))], 'rater':['1']*len(mr_cyl_1_corr)+['2']*len(mr_cyl_2_corr)+['3']*len(mr_cyl_3_corr)+['DLC']*len(footfault_FI_2d_corr), 'rating':np.concatenate((mr_cyl_1_corr, mr_cyl_2_corr, mr_cyl_3_corr, footfault_FI_2d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_1_corr))]+[i for i in range(len(mr_cyl_2_corr))]+[i for i in range(len(mr_cyl_3_corr))]+[i for i in range(len(footfault_3d_corr))], 'rater':['1']*len(mr_cyl_1_corr)+['2']*len(mr_cyl_2_corr)+['3']*len(mr_cyl_3_corr)+['DLC']*len(footfault_3d_corr), 'rating':np.concatenate((mr_cyl_1_corr, mr_cyl_2_corr, mr_cyl_3_corr, footfault_3d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_1_corr))]+[i for i in range(len(mr_cyl_2_corr))]+[i for i in range(len(mr_cyl_3_corr))]+[i for i in range(len(footfault_FI_3d_corr))], 'rater':['1']*len(mr_cyl_1_corr)+['2']*len(mr_cyl_2_corr)+['3']*len(mr_cyl_3_corr)+['DLC']*len(footfault_FI_3d_corr), 'rating':np.concatenate((mr_cyl_1_corr, mr_cyl_2_corr, mr_cyl_3_corr, footfault_FI_3d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])

# ============================================================================================================================================


## Correlation Feature Cylinder Touch within MR and between MR and DLC plus FI 2D and 3D
# ============================================================================================================================================
mr_cyl_touch_1=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Jan_touch.csv", index_col=0, delimiter=',', header=None)
mr_cyl_touch_2=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Nicole_touch.csv", index_col=0, delimiter=',', header=None)
mr_cyl_touch_3=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Jule_touch.csv", index_col=0, delimiter=',', header=None)

patients=['PB_T2_3_1', 'PB_T2_3_2', 'PB_T2_3_3', 'PB_T2_4_2', 'PB_T2_5_1', 'PB_T2_6_1', 'PB_T2_6_2', 'PB_T3_23_1', 'PB_T3_23_2', 'PB_T3_23_3', 'PB_T3_24_1', 'PB_T3_24_2', 'PB_T3_24_3']

max_i=np.array([np.nanmax(mr_cyl_touch_1.loc['PB_T2_3_1']), np.nanmax(mr_cyl_touch_2.loc['PB_T2_3_1']), np.nanmax(mr_cyl_touch_3.loc['PB_T2_3_1']), np.nanmax(touch_2d.loc['PB_T2_3_1']), np.nanmax(touch_FI_2d.loc['PB_T2_3_1']), np.nanmax(touch_3d.loc['PB_T2_3_1']), np.nanmax(touch_FI_3d.loc['PB_T2_3_1'])])
mr_cyl_touch_1_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
mr_cyl_touch_2_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
mr_cyl_touch_3_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
touch_2d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
touch_FI_2d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
touch_3d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
touch_FI_3d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))

for i in mr_cyl_touch_1.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_touch_1_corr[int(i-15):int(i+16)]=1
for i in mr_cyl_touch_2.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_touch_2_corr[int(i-15):int(i+16)]=1
for i in mr_cyl_touch_3.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_touch_3_corr[int(i-15):int(i+16)]=1
for i in touch_2d.loc['PB_T2_3_1']:
    if i>0:
        touch_2d_corr[int(i-15):int(i+16)]=1
for i in touch_FI_2d.loc['PB_T2_3_1']:
    if i>0:
        touch_FI_2d_corr[int(i-15):int(i+16)]=1
for i in touch_3d.loc['PB_T2_3_1']:
    if i>0:
        touch_3d_corr[int(i-15):int(i+16)]=1
for i in touch_FI_3d.loc['PB_T2_3_1']:
    if i>0:
        touch_FI_3d_corr[int(i-15):int(i+16)]=1
    
for z in patients[1:]:
    max_i=np.array([np.nanmax(mr_cyl_touch_1.loc[z]), np.nanmax(mr_cyl_touch_2.loc[z]), np.nanmax(mr_cyl_touch_3.loc[z]), np.nanmax(touch_2d.loc[z]), np.nanmax(touch_FI_2d.loc[z]), np.nanmax(touch_3d.loc[z]), np.nanmax(touch_FI_3d.loc[z])])
    if all(max_i==max_i):
        mr_cyl_touch_1_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        mr_cyl_touch_2_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        mr_cyl_touch_3_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        touch_2d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        touch_FI_2d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        touch_3d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        touch_FI_3d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        for i in mr_cyl_touch_1.loc[z]:
            if i>0:
                mr_cyl_touch_1_corr[int(i-15):int(i+16)]=1
        for i in mr_cyl_touch_2.loc[z]:
            if i>0:
                mr_cyl_touch_2_corr[int(i-15):int(i+16)]=1
        for i in mr_cyl_touch_3.loc[z]:
            if i>0:
                mr_cyl_touch_3_corr[int(i-15):int(i+16)]=1
        for i in touch_2d.loc[z]:
            if i>0:
                touch_2d_corr[int(i-15):int(i+16)]=1
        for i in touch_FI_2d.loc[z]:
            if i>0:
                touch_FI_2d_corr[int(i-15):int(i+16)]=1
        for i in touch_3d.loc[z]:
            if i>0:
                touch_3d_corr[int(i-15):int(i+16)]=1
        for i in touch_FI_3d.loc[z]:
            if i>0:
                touch_FI_3d_corr[int(i-15):int(i+16)]=1
        mr_cyl_touch_1_corr=np.r_[mr_cyl_touch_1_corr, mr_cyl_touch_1_corr_temp]
        mr_cyl_touch_2_corr=np.r_[mr_cyl_touch_2_corr, mr_cyl_touch_2_corr_temp]
        mr_cyl_touch_3_corr=np.r_[mr_cyl_touch_3_corr, mr_cyl_touch_3_corr_temp]
        touch_2d_corr=np.r_[touch_2d_corr, touch_2d_corr_temp]
        touch_FI_2d_corr=np.r_[touch_FI_2d_corr, touch_FI_2d_corr_temp]
        touch_3d_corr=np.r_[touch_3d_corr, touch_3d_corr_temp]
        touch_FI_3d_corr=np.r_[touch_FI_3d_corr, touch_FI_3d_corr_temp]

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), 1, jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(3, 3))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_MR_CY_touch.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_1_corr, touch_2d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), 1, jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, touch_2d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), 1, jaccard(mr_cyl_touch_3_corr, touch_2d_corr)],
                       [jaccard(mr_cyl_touch_1_corr, touch_2d_corr), jaccard(mr_cyl_touch_2_corr, touch_2d_corr), jaccard(mr_cyl_touch_3_corr, touch_2d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_touch_2d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_1_corr, touch_FI_2d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), 1, jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, touch_FI_2d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), 1, jaccard(mr_cyl_touch_3_corr, touch_FI_2d_corr)],
                       [jaccard(mr_cyl_touch_1_corr, touch_FI_2d_corr), jaccard(mr_cyl_touch_2_corr, touch_FI_2d_corr), jaccard(mr_cyl_touch_3_corr, touch_FI_2d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_touch_FI_2d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_1_corr, touch_3d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), 1, jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, touch_3d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), 1, jaccard(mr_cyl_touch_3_corr, touch_3d_corr)],
                       [jaccard(mr_cyl_touch_1_corr, touch_3d_corr), jaccard(mr_cyl_touch_2_corr, touch_3d_corr), jaccard(mr_cyl_touch_3_corr, touch_3d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_touch_3d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_1_corr, touch_FI_3d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_2_corr), 1, jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, touch_FI_3d_corr)], 
                       [jaccard(mr_cyl_touch_1_corr, mr_cyl_touch_3_corr), jaccard(mr_cyl_touch_2_corr, mr_cyl_touch_3_corr), 1, jaccard(mr_cyl_touch_3_corr, touch_FI_3d_corr)],
                       [jaccard(mr_cyl_touch_1_corr, touch_FI_3d_corr), jaccard(mr_cyl_touch_2_corr, touch_FI_3d_corr), jaccard(mr_cyl_touch_3_corr, touch_FI_3d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_touch_FI_3d.png", format='png', dpi=200)



print('CY touch 3D')
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_touch_1_corr))]+[i for i in range(len(mr_cyl_touch_2_corr))]+[i for i in range(len(mr_cyl_touch_3_corr))], 'rater':['1']*len(mr_cyl_touch_1_corr)+['2']*len(mr_cyl_touch_2_corr)+['3']*len(mr_cyl_touch_3_corr), 'rating':np.concatenate((mr_cyl_touch_1_corr, mr_cyl_touch_2_corr, mr_cyl_touch_3_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_touch_1_corr))]+[i for i in range(len(mr_cyl_touch_2_corr))]+[i for i in range(len(mr_cyl_touch_3_corr))]+[i for i in range(len(touch_2d_corr))], 'rater':['1']*len(mr_cyl_touch_1_corr)+['2']*len(mr_cyl_touch_2_corr)+['3']*len(mr_cyl_touch_3_corr)+['DLC']*len(touch_2d_corr), 'rating':np.concatenate((mr_cyl_touch_1_corr, mr_cyl_touch_2_corr, mr_cyl_touch_3_corr, touch_2d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_touch_1_corr))]+[i for i in range(len(mr_cyl_touch_2_corr))]+[i for i in range(len(mr_cyl_touch_3_corr))]+[i for i in range(len(touch_FI_2d_corr))], 'rater':['1']*len(mr_cyl_touch_1_corr)+['2']*len(mr_cyl_touch_2_corr)+['3']*len(mr_cyl_touch_3_corr)+['DLC']*len(touch_FI_2d_corr), 'rating':np.concatenate((mr_cyl_touch_1_corr, mr_cyl_touch_2_corr, mr_cyl_touch_3_corr, touch_FI_2d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_touch_1_corr))]+[i for i in range(len(mr_cyl_touch_2_corr))]+[i for i in range(len(mr_cyl_touch_3_corr))]+[i for i in range(len(touch_3d_corr))], 'rater':['1']*len(mr_cyl_touch_1_corr)+['2']*len(mr_cyl_touch_2_corr)+['3']*len(mr_cyl_touch_3_corr)+['DLC']*len(touch_3d_corr), 'rating':np.concatenate((mr_cyl_touch_1_corr, mr_cyl_touch_2_corr, mr_cyl_touch_3_corr, touch_3d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_touch_1_corr))]+[i for i in range(len(mr_cyl_touch_2_corr))]+[i for i in range(len(mr_cyl_touch_3_corr))]+[i for i in range(len(touch_FI_3d_corr))], 'rater':['1']*len(mr_cyl_touch_1_corr)+['2']*len(mr_cyl_touch_2_corr)+['3']*len(mr_cyl_touch_3_corr)+['DLC']*len(touch_FI_3d_corr), 'rating':np.concatenate((mr_cyl_touch_1_corr, mr_cyl_touch_2_corr, mr_cyl_touch_3_corr, touch_FI_3d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])

# ============================================================================================================================================


## Correlation Feature Cylinder Drag within MR and between MR and DLC plus FI 2D and 3D
# ============================================================================================================================================
mr_cyl_drag_1=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Jan_drag.csv", index_col=0, delimiter=',', header=None)
mr_cyl_drag_2=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Nicole_drag.csv", index_col=0, delimiter=',', header=None)
mr_cyl_drag_3=pd.read_csv("/home/user/owncloud/3D_videos/manual_raters/Manual_Cylinder_Analyse_Jule_drag.csv", index_col=0, delimiter=',', header=None)

patients=['PB_T2_3_1', 'PB_T2_3_2', 'PB_T2_3_3', 'PB_T2_4_2', 'PB_T2_5_1', 'PB_T2_6_1', 'PB_T2_6_2', 'PB_T3_23_1', 'PB_T3_23_2', 'PB_T3_23_3', 'PB_T3_24_1', 'PB_T3_24_2', 'PB_T3_24_3']

max_i=np.array([np.nanmax(mr_cyl_drag_1.loc['PB_T2_3_1']), np.nanmax(mr_cyl_drag_2.loc['PB_T2_3_1']), np.nanmax(mr_cyl_drag_3.loc['PB_T2_3_1']), np.nanmax(drag_2d.loc['PB_T2_3_1']), np.nanmax(drag_FI_2d.loc['PB_T2_3_1']), np.nanmax(drag_3d.loc['PB_T2_3_1']), np.nanmax(drag_FI_3d.loc['PB_T2_3_1'])])
mr_cyl_drag_1_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
mr_cyl_drag_2_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
mr_cyl_drag_3_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
drag_2d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
drag_FI_2d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
drag_3d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))
drag_FI_3d_corr=np.zeros(shape=(int(np.nanmax(max_i)+11),))

for i in mr_cyl_drag_1.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_drag_1_corr[int(i-15):int(i+16)]=1
for i in mr_cyl_drag_2.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_drag_2_corr[int(i-15):int(i+16)]=1
for i in mr_cyl_drag_3.loc['PB_T2_3_1']:
    if i>0:
        mr_cyl_drag_3_corr[int(i-15):int(i+16)]=1
for i in drag_2d.loc['PB_T2_3_1']:
    if i>0:
        drag_2d_corr[int(i-15):int(i+16)]=1
for i in drag_FI_2d.loc['PB_T2_3_1']:
    if i>0:
        drag_FI_2d_corr[int(i-15):int(i+16)]=1
for i in drag_3d.loc['PB_T2_3_1']:
    if i>0:
        drag_3d_corr[int(i-15):int(i+16)]=1
for i in drag_FI_3d.loc['PB_T2_3_1']:
    if i>0:
        drag_FI_3d_corr[int(i-15):int(i+16)]=1
    
for z in patients[1:]:
    max_i=np.array([np.nanmax(mr_cyl_drag_1.loc[z]), np.nanmax(mr_cyl_drag_2.loc[z]), np.nanmax(mr_cyl_drag_3.loc[z]), np.nanmax(drag_2d.loc[z]), np.nanmax(drag_FI_2d.loc[z]), np.nanmax(drag_3d.loc[z]), np.nanmax(drag_FI_3d.loc[z])])
    if all(max_i==max_i):
        mr_cyl_drag_1_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        mr_cyl_drag_2_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        mr_cyl_drag_3_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        drag_2d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        drag_FI_2d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        drag_3d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        drag_FI_3d_corr_temp=np.zeros(shape=(int(np.nanmax(max_i)+11),))
        for i in mr_cyl_drag_1.loc[z]:
            if i>0:
                mr_cyl_drag_1_corr[int(i-15):int(i+16)]=1
        for i in mr_cyl_drag_2.loc[z]:
            if i>0:
                mr_cyl_drag_2_corr[int(i-15):int(i+16)]=1
        for i in mr_cyl_drag_3.loc[z]:
            if i>0:
                mr_cyl_drag_3_corr[int(i-15):int(i+16)]=1
        for i in drag_2d.loc[z]:
            if i>0:
                drag_2d_corr[int(i-15):int(i+16)]=1
        for i in drag_FI_2d.loc[z]:
            if i>0:
                drag_FI_2d_corr[int(i-15):int(i+16)]=1
        for i in drag_3d.loc[z]:
            if i>0:
                drag_3d_corr[int(i-15):int(i+16)]=1
        for i in drag_FI_3d.loc[z]:
            if i>0:
                drag_FI_3d_corr[int(i-15):int(i+16)]=1
        mr_cyl_drag_1_corr=np.r_[mr_cyl_drag_1_corr, mr_cyl_drag_1_corr_temp]
        mr_cyl_drag_2_corr=np.r_[mr_cyl_drag_2_corr, mr_cyl_drag_2_corr_temp]
        mr_cyl_drag_3_corr=np.r_[mr_cyl_drag_3_corr, mr_cyl_drag_3_corr_temp]
        drag_2d_corr=np.r_[drag_2d_corr, drag_2d_corr_temp]
        drag_FI_2d_corr=np.r_[drag_FI_2d_corr, drag_FI_2d_corr_temp]
        drag_3d_corr=np.r_[drag_3d_corr, drag_3d_corr_temp]
        drag_FI_3d_corr=np.r_[drag_FI_3d_corr, drag_FI_3d_corr_temp]

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), 1, jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(3, 3))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_MR_CY_drag.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_1_corr, drag_2d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), 1, jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, drag_2d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), 1, jaccard(mr_cyl_drag_3_corr, drag_2d_corr)],
                       [jaccard(mr_cyl_drag_1_corr, drag_2d_corr), jaccard(mr_cyl_drag_2_corr, drag_2d_corr), jaccard(mr_cyl_drag_3_corr, drag_2d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_drag_2d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_1_corr, drag_FI_2d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), 1, jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, drag_FI_2d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), 1, jaccard(mr_cyl_drag_3_corr, drag_FI_2d_corr)],
                       [jaccard(mr_cyl_drag_1_corr, drag_FI_2d_corr), jaccard(mr_cyl_drag_2_corr, drag_FI_2d_corr), jaccard(mr_cyl_drag_3_corr, drag_FI_2d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_drag_FI_2d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_1_corr, drag_3d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), 1, jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, drag_3d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), 1, jaccard(mr_cyl_drag_3_corr, drag_3d_corr)],
                       [jaccard(mr_cyl_drag_1_corr, drag_3d_corr), jaccard(mr_cyl_drag_2_corr, drag_3d_corr), jaccard(mr_cyl_drag_3_corr, drag_3d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_drag_3d.png", format='png', dpi=200)

sn.set_theme(style="white")
d = pd.DataFrame(data=np.array([[1,jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_1_corr, drag_FI_3d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_2_corr), 1, jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, drag_FI_3d_corr)], 
                       [jaccard(mr_cyl_drag_1_corr, mr_cyl_drag_3_corr), jaccard(mr_cyl_drag_2_corr, mr_cyl_drag_3_corr), 1, jaccard(mr_cyl_drag_3_corr, drag_FI_3d_corr)],
                       [jaccard(mr_cyl_drag_1_corr, drag_FI_3d_corr), jaccard(mr_cyl_drag_2_corr, drag_FI_3d_corr), jaccard(mr_cyl_drag_3_corr, drag_FI_3d_corr), 1]])
                                        , columns=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'], index=['Manual Rater 1', 'Manual Rater 2', 'Manual Rater 3', 'DLC'])

mask = np.tril(np.ones_like(d, dtype=bool), k=-1)
f, ax = plt.subplots(figsize=(4, 4))
cmap = sn.diverging_palette(10, 250, s=90, l=40, as_cmap=True)
sn.set(font_scale=3)
hmap=sn.heatmap(d, mask=mask.T, cmap=cmap, vmax=1, vmin=0, center=0, annot=True, square=True, linewidths=1, cbar_kws={"shrink": .6}, fmt=".3g")
hmap.set_xticklabels(hmap.get_xmajorticklabels(), fontsize = 40, rotation=45)
hmap.set_yticklabels(hmap.get_ymajorticklabels(), fontsize = 40)
hmap.figure.set_size_inches(22, 18)
plt.tight_layout()
hmap.figure.savefig("/home/user/owncloud/thesis_figures/Correlation_DLC_CY_drag_FI_3d.png", format='png', dpi=200)



print('CY drag 3D')
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_drag_1_corr))]+[i for i in range(len(mr_cyl_drag_2_corr))]+[i for i in range(len(mr_cyl_drag_3_corr))], 'rater':['1']*len(mr_cyl_drag_1_corr)+['2']*len(mr_cyl_drag_2_corr)+['3']*len(mr_cyl_drag_3_corr), 'rating':np.concatenate((mr_cyl_drag_1_corr, mr_cyl_drag_2_corr, mr_cyl_drag_3_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_drag_1_corr))]+[i for i in range(len(mr_cyl_drag_2_corr))]+[i for i in range(len(mr_cyl_drag_3_corr))]+[i for i in range(len(drag_2d_corr))], 'rater':['1']*len(mr_cyl_drag_1_corr)+['2']*len(mr_cyl_drag_2_corr)+['3']*len(mr_cyl_drag_3_corr)+['DLC']*len(drag_2d_corr), 'rating':np.concatenate((mr_cyl_drag_1_corr, mr_cyl_drag_2_corr, mr_cyl_drag_3_corr, drag_2d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_drag_1_corr))]+[i for i in range(len(mr_cyl_drag_2_corr))]+[i for i in range(len(mr_cyl_drag_3_corr))]+[i for i in range(len(drag_FI_2d_corr))], 'rater':['1']*len(mr_cyl_drag_1_corr)+['2']*len(mr_cyl_drag_2_corr)+['3']*len(mr_cyl_drag_3_corr)+['DLC']*len(drag_FI_2d_corr), 'rating':np.concatenate((mr_cyl_drag_1_corr, mr_cyl_drag_2_corr, mr_cyl_drag_3_corr, drag_FI_2d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_drag_1_corr))]+[i for i in range(len(mr_cyl_drag_2_corr))]+[i for i in range(len(mr_cyl_drag_3_corr))]+[i for i in range(len(drag_3d_corr))], 'rater':['1']*len(mr_cyl_drag_1_corr)+['2']*len(mr_cyl_drag_2_corr)+['3']*len(mr_cyl_drag_3_corr)+['DLC']*len(drag_3d_corr), 'rating':np.concatenate((mr_cyl_drag_1_corr, mr_cyl_drag_2_corr, mr_cyl_drag_3_corr, drag_3d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])
df=pd.DataFrame({'index':[i for i in range(len(mr_cyl_drag_1_corr))]+[i for i in range(len(mr_cyl_drag_2_corr))]+[i for i in range(len(mr_cyl_drag_3_corr))]+[i for i in range(len(drag_FI_3d_corr))], 'rater':['1']*len(mr_cyl_drag_1_corr)+['2']*len(mr_cyl_drag_2_corr)+['3']*len(mr_cyl_drag_3_corr)+['DLC']*len(drag_FI_3d_corr), 'rating':np.concatenate((mr_cyl_drag_1_corr, mr_cyl_drag_2_corr, mr_cyl_drag_3_corr, drag_FI_3d_corr))})
icc = pg.intraclass_corr(data=df, targets='index', raters='rater', ratings='rating', nan_policy='omit')
print(icc.loc[2])

# ============================================================================================================================================