Untitled

import pandas as pd
import numpy as np
from sklearn.metrics import recall_score
from bokeh.plotting import figure, show
from bokeh.io import output_notebook
from bokeh.models import Label

# Load data from Excel sheet 'A'
file_path = 'your_file.xlsx'  # Replace with your Excel file path
df = pd.read_excel(file_path, sheet_name='A')

# Replace 'true_labels_column' and 'pred_prob_model_1' with actual column names
y_true = df['true_labels_column']  # True labels
y_pred_prob_model_1 = df['pred_prob_model_1']  # Predicted probabilities from Model 1

# Filter data for positive instances (where true label is 1)
positive_indices = y_true == 1
y_true_positive = y_true[positive_indices]
y_pred_prob_positive = y_pred_prob_model_1[positive_indices]

# Define thresholds from 0 to 1
thresholds = np.arange(0, 1.05, 0.05)
recalls = []

# Calculate recall for each threshold
for threshold in thresholds:
    # Binarize predictions based on threshold
    y_pred = (y_pred_prob_positive >= threshold).astype(int)
    
    # Only compute recall for positive instances (where true label is 1)
    recalls.append(recall_score(y_true_positive, y_pred))

# Output to notebook (or use output_file('filename.html') for a standalone HTML file)
output_notebook()

# Create a Bokeh figure
p = figure(title='Threshold vs. Recall (Positive Class Only)',
           x_axis_label='Threshold',
           y_axis_label='Recall',
           width=900, height=500,
           toolbar_location=None)

# Plot Recall vs. Threshold
p.line(thresholds, recalls, line_color='blue', line_width=2, alpha=0.8, legend_label='Recall')

# Customize title
p.title.align = 'center'
p.title.text_font_size = '18pt'
p.title.text_font = 'bold'

# Customize legend
p.legend.location = 'top_left'
p.legend.label_text_font_size = '12pt'

# Add annotations for maximum recall
max_recall = max(recalls)
max_threshold = thresholds[recalls.index(max_recall)]
p.add_layout(Label(x=max_threshold, y=max_recall, 
                   text=f'Max Recall: {max_recall:.2f} at Threshold: {max_threshold:.2f}', 
                   text_color='red', text_font_size='12pt'))

# Customize grid
p.xgrid.grid_line_color = 'lightgray'
p.ygrid.grid_line_color = 'lightgray'

# Customize axis labels
p.xaxis.axis_label_text_font_size = '14pt'
p.yaxis.axis_label_text_font_size = '14pt'
p.xaxis.major_label_text_font_size = '12pt'
p.yaxis.major_label_text_font_size = '12pt'

# Show the plot
show(p)