mail@pastecode.io avatar
7 months ago
3.5 kB
# Preprocess data
# Split data into input and output variables
X = df[['tenant_id', 'ticket_desc', 'ticket_summary']]
y = df[['ticket_call_type', 'ticket_type', 'ticket_category', 'ticket_type', 'ticket_item', 'ticket_severity']]

# Convert text data to numeric features using CountVectorizer
X['text'] = X['ticket_desc'] + ' ' + X['ticket_summary']
X_text = X['text']
cv = CountVectorizer(stop_words='english',max_features=1000)
X_cv = cv.fit_transform(X_text)
X_cv_df = pd.DataFrame(X_cv.toarray(), columns=cv.get_feature_names_out())
X_cv_df = X_cv_df.values.reshape(X.shape[0], -1)

# Split data into training and test sets
X_train, X_test, y_train, y_test = train_test_split(X_cv_df, y, test_size=0.2, random_state=42)

# Train random forest classifier to predict output variables
rf = RandomForestClassifier(n_estimators=100, random_state=42)
rf.fit(X_train, y_train)

from chatterbot.trainers import ChatterBotCorpusTrainer
from chatterbot import ChatBot

import nltk
import ssl

    _create_unverified_https_context = ssl._create_unverified_context
except AttributeError:
    ssl._create_default_https_context = _create_unverified_https_context

# Define chatbot
chatbot = ChatBot('TicketBot')

trainer = ChatterBotCorpusTrainer(chatbot)

# Train chatbot using corpus

# Define function to extract tenant name from user input
def extract_tenant(input_text):
    tenant_names = ['INfra', 'Tool_Team', 'Sodexo_Core', 'Platform_Team']
    input_text = input_text.lower()
    for name in tenant_names:
        if name.lower() in input_text:
            return name
    return None

# Define function to generate incident/ticket based on user input
def generate_ticket(input_text):
    # Extract tenant name from user input
    tenant = extract_tenant(input_text)
    if not tenant:
        return 'Please provide a valid tenant name'
    # Get description and summary from user input
    input_text = input_text.lower()
    description = input_text.split('description:')[1].strip() if 'description:' in input_text else ''
    summary = input_text.split('summary:')[1].strip() if 'summary:' in input_text else ''
    # Use machine learning model to predict output variables
    input_vector = cv.transform(np.array([tenant, description, summary]))
    input_vector = input_vector.toarray().reshape(1, -1)
    prediction = rf.predict(input_vector)
    # Create incident/ticket
    ticket = {
        'Tenant': tenant,
        'Description': description,
        'Summary': summary,
        'Call_type': prediction[0][0],
        'Ticket_Type': prediction[0][1],
        'Requestor': prediction[0][2],
        'Category': prediction[0][3],
        'Type': prediction[0][4],
        'Item': prediction[0][5],
        'Severity': prediction[0][6]
    return ticket

# Define function to handle user input and generate response
def handle_input(input_text):
    # Generate incident/ticket
    ticket = generate_ticket(input_text)
    # Auto-populate ticket fields
    ticket_type = ticket['ticket_type']
    requestor = ticket['requestor']
    category = ticket['category']
    ticket_item = ticket['ticket_item']
    severity = ticket['severity']
    description = ticket['description']
    tenant = ticket['tenant']
    call_type = ticket['call_type']
    summary = ticket['summary']