Ever wondered what happens after you raise an IT support ticket?
Behind every ticket, there's a trail of updates, reassignments, priority changes, delays, and finally... a resolution. Now imagine that happening more than 141,000 times.
That's exactly what this project explores.
Using real IT incident logs from a public ServiceNow dataset, I cleaned messy event-level data, converted multiple updates into unique incidents, explored how different factors affect SLA performance, trained machine learning models to predict SLA breaches, ranked incidents based on risk, and even estimated future workload using time-series forecasting.
Instead of just building a prediction model and calling it a day, I wanted the entire workflow to feel useful. That's why the final results are presented through interactive Power BI dashboards that make it easy to spot trends, monitor performance, and quickly identify incidents that need attention.
- Data cleaning and preprocessing
- Exploratory Data Analysis (EDA)
- SLA breach prediction using Machine Learning
- Risk scoring for incident prioritization
- 30-day incident forecasting
- Interactive Power BI dashboards
- Project report
Python • SQL • Pandas • Scikit-learn • Facebook Prophet • Power BI
IT Incident Log Dataset (Public)
141,712 event logs
24,918 unique incidents
Anonymized data extracted from a real ServiceNow environment.
🔗 https://www.kaggle.com/datasets/shamiulislamshifat/it-incident-log-dataset
Dashboard screenshots are included in this repository.
₊˚⊹ If you have any ideas to improve the project or just want to discuss data analytics, feel free to reach out. Always happy to connect :)