app.py

import gradio as gr
import pandas as pd
import pickle
import shap
import xgboost
import matplotlib
import matplotlib.pyplot as plt
import io
matplotlib.use('Agg')
pc = ["Enterprise System", "Management Information System", "Safety Critical System", "Transaction Processing System" ]
rc = ['Constraints','Functional','Interfaces','Performance','Reliability&Availability','Safety','Security','Standards','Supportability','Usability']
rtc = ['Budget','Business','Cost','Design','Functional Validity','Organizational Environment','Overdrawn Budget','People','Performance','Personal','Planning&Control','Process','Project complexity','Quality','Requirement','Resource availability','Schedule','Software','Team','TimeDimension','Unrealistic Requirements','User']
mor = ['Extreme','High','Low','Medium','Negligible','Very High','Very Low']
imp = ["Low", "Catastrophic", "High", "Insignificant", "High", "Moderate"]
dor = ['Cost','Estimations','Organizational Environment','Organizational Requirements','Planning and Control','Project Complexity','Project complexity','Requirements','Schedule','Software Requirement','Team','User','Planning and Control']
def ml_function(probability, affecting_no_of_modules,
                fixing_duration, project_category, requirement_category, risk_target_category,
                magnitude_of_risk, impact, dimension_of_risk, fix_cost_percent_of_project):
    data = list()
    data.append(probability)
    data.append(affecting_no_of_modules)
    data.append(fixing_duration)
    data.extend([bool(c in project_category) for c in pc])
    data.extend([bool(c in requirement_category) for c in rc])
    data.extend([bool(c in risk_target_category) for c in rtc])
    data.extend([bool(c in magnitude_of_risk) for c in mor])
    data.extend([bool(c in impact) for c in imp])
    data.extend([bool(c in dimension_of_risk) for c in dor])
    fix_cost_percent_of_project_one_hot = [False] * 10
    fix_cost_percent_of_project_one_hot[fix_cost_percent_of_project] = True
    data.extend(fix_cost_percent_of_project_one_hot)

    df = pd.DataFrame([data])
    df.columns = ['Probability', 'Afftecting No of Modules', 'Fixing Duration (Days)',
       'project Category_Enterprise System',
       'project Category_Management Information System',
       'project Category_Safety Critical System',
       'project Category_Transaction Processing System',
       'Requirement Category_Constraints', 'Requirement Category_Functional',
       'Requirement Category_Interfaces', 'Requirement Category_Performance',
       'Requirement Category_Reliability & Availability',
       'Requirement Category_Safety', 'Requirement Category_Security',
       'Requirement Category_Standards', 'Requirement Category_Supportability',
       'Requirement Category_Usability', 'Risk Target Category_Budget',
       'Risk Target Category_Business', 'Risk Target Category_Cost',
       'Risk Target Category_Design',
       'Risk Target Category_FunctionalValidity',
       'Risk Target Category_Organizational Environment',
       'Risk Target Category_Overdrawn Budget', 'Risk Target Category_People',
       'Risk Target Category_Performance', 'Risk Target Category_Personal',
       'Risk Target Category_Planning & Control',
       'Risk Target Category_Process',
       'Risk Target Category_Project complexity',
       'Risk Target Category_Quality', 'Risk Target Category_Requirement',
       'Risk Target Category_Resource availability',
       'Risk Target Category_Schedule', 'Risk Target Category_Software',
       'Risk Target Category_Team', 'Risk Target Category_Time Dimension',
       'Risk Target Category_Unrealistic Requirements',
       'Risk Target Category_User', 'Magnitude of Risk_Extreme',
       'Magnitude of Risk_High', 'Magnitude of Risk_Low',
       'Magnitude of Risk_Medium', 'Magnitude of Risk_Negligible',
       'Magnitude of Risk_Very High', 'Magnitude of Risk_Very Low',
       'Impact_Low', 'Impact_catastrophic', 'Impact_high',
       'Impact_insignificant', 'Impact_moderate', 'Dimension of Risk_Cost',
       'Dimension of Risk_Estimations',
       'Dimension of Risk_Organizational Environment',
       'Dimension of Risk_Organizational Requirements',
       'Dimension of Risk_Planning and Control',
       'Dimension of Risk_Project Complexity',
       'Dimension of Risk_Project complexity',
       'Dimension of Risk_Requirements', 'Dimension of Risk_Schedule',
       'Dimension of Risk_Software Requirement', 'Dimension of Risk_Team',
       'Dimension of Risk_User', 'Dimension of Risk_planning and control',
       'Fix Cost (\% of Project)_0', 'Fix Cost (\% of Project)_1',
       'Fix Cost (\% of Project)_10', 'Fix Cost (\% of Project)_11',
       'Fix Cost (\% of Project)_2', 'Fix Cost (\% of Project)_21',
       'Fix Cost (\% of Project)_3', 'Fix Cost (\% of Project)_4',
       'Fix Cost (\% of Project)_5', 'Fix Cost (\% of Project)_6',
       'Fix Cost (\% of Project)_?']
    df = df.reset_index(drop=True)
    for col in df.columns:
        if df[col].dtype == 'object':
            df[col] = pd.to_numeric(df[col])
    xgb_model_loaded = pickle.load(open('xgboost_project_risk_prediction.pickle', "rb"))
    pred = xgb_model_loaded.predict(df[:1])
    return str(pred[0])

# Define the Gradio interface with appropriate input components
iface = gr.Interface(
    fn=ml_function,
    inputs=[
        gr.Number(label="Probability"),
        gr.Number(label="Affecting No of Modules"),
        gr.Number(label="Fixing Duration (Days)"),
        gr.Radio(choices=pc, label="Project Category"),
        gr.Radio(choices=rc, label="Requirement Category"),
        gr.Radio(choices=rtc, label="Risk Target Category"),
        gr.Radio(choices=mor, label="Magnitude of Risk"),
        gr.Radio(choices=imp, label="Impact"),
        gr.Radio(choices=dor, label="Dimension of Risk"),
        gr.Slider(minimum=0, maximum=10,step=1, label="Fix Cost (% of Project)")
    ],
    outputs=[
        gr.Label(label="Risk Level")
    ]
)

# Run the Gradio app
iface.launch()