Outlier_detection_PO

import pandas as pd
from sklearn.neighbors import LocalOutlierFactor

# Load purchase order data into a pandas DataFrame
data = pd.read_csv('purchase_order_data.csv')

# Preprocess the data if needed (e.g., handle missing values, adjust data types)

# Select features for anomaly detection
features = ['amount', 'vendor', 'date']  # Adjust the column names according to your data

# Create a new DataFrame with only the selected features
data_selected = data[features]

# Create a Local Outlier Factor model
lof_model = LocalOutlierFactor(contamination=0.05)  # Adjust the contamination parameter based on your requirements

# Fit the model to the selected data
lof_model.fit_predict(data_selected)

# Obtain the anomaly scores
anomaly_scores = lof_model.negative_outlier_factor_

# Add the anomaly scores as a new column in the original DataFrame
data['anomaly_score'] = anomaly_scores

# Analyze the anomalies (e.g., identify and investigate the data points with low anomaly scores)

# Once the model is trained and anomalies are identified, you can integrate the anomaly detection system into your ERP accounting system to flag and handle potential anomalies in purchase orders.