ABC Tech ITSM Incident Analysis & Predictive Automation

Project Overview

project UI : https://itsm-project-ui.vercel.app/

This project delivers a robust, end-to-end data science pipeline built to analyze historical IT Service Management (ITSM) incident data from a major tech firm, ABC Tech, and deploy predictive models for operational optimization.

The primary objective was to move ITSM operations from reactive to proactive by automating triage and forecasting future resource needs.

---

✨ Key Achievements & Deliverables

The project successfully addressed all four key client objectives, resulting in quantifiable improvements to efficiency and risk management. The project utilized a 45,000+ record, SQL-sourced ITSM dataset.

Goal	Description	Key Achievement
1. High Priority Prediction	Forecast whether an incoming ticket will be high priority (P1/P2).	96% Accuracy and 0.89 AUC achieved by resolving critical data leakage.
2. Incident Forecasting	Forecast incident volume across quarterly and annual horizons.	Stabilized forecasting models (Exponential Smoothing, Regression) to provide reliable, non-negative volume projections for staffing and capacity planning.
3. Auto-Tagging	Automatically assign the correct Priority (P1-P5) and Department (CI_Cat).	75% Accuracy in multi-class Department Tagging and high-confidence Priority routing.
4. RFC Failure Prediction	Predict the likelihood of a Request-for-Change (RFC) leading to a failure or misconfiguration.	98% Test F1-Score achieved using a sophisticated XGBoost model with SMOTE, enabling proactive risk mitigation.

---

⚙️ Technical Stack

• Language: Python (3.x)
• Data Handling: Pandas, NumPy, MySQL Connector
• Machine Learning: Scikit-learn (Pipelines, ColumnTransformer, RandomForest), XGBoost
• Time Series: statsmodels (Exponential Smoothing, Holt), Scikit-learn (Linear Regression)
• Imbalance Handling: imblearn (SMOTE)
• Visualization: Matplotlib, Seaborn
• Deployment Assets: joblib (for saving trained models and scalers)

---

🛠️ Installation and Setup

1. Clone the repository:

   git clone [Your Repository URL]
   cd abc-tech-itsm-analysis

2. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Linux/Mac
   .\venv\Scripts\activate    # On Windows

3. Install dependencies:

   pip install pandas numpy scikit-learn xgboost matplotlib seaborn mysql-connector-python imbalanced-learn statsmodels

4. Database Connection: The notebook connects to a MySQL database using internal credentials. If you are running this locally, ensure you update the database connection details in the Data Acquisition & Preprocessing cell of the notebook.

---

🚀 Execution

The core of the project is contained within the ABC_TECH_ITSM_Project.ipynb notebook.

1. Open the Jupyter Notebook:

   jupyter notebook

2. Run all cells sequentially. The notebook executes the following steps:
   • Preprocessing: Robust data cleaning, imputation, scaling, and one-hot encoding.
   • Goal 1: High Priority Incident Prediction.
   • Goal 2: Incident Volume Forecasting (Quarterly, Annual, Monthly).
   • Goal 3: Department Auto-Tagging.
   • Goal 4: RFC Failure Prediction.