Data Mining Approaches: Tasks, CRISP-DM & Methodologies

Approaches in Data Mining

Key Methods, Tasks, and Processes like CRISP-DM

Approaches in Data Mining

Key Methods, Tasks, and Processes like CRISP-DM

---

Photo by Deng Xiang on Unsplash

• Introduction to Data Mining
• Key Data Mining Tasks
• CRISP-DM: Standard Process Model
• Other Data Mining Methodologies
• Conclusion

---

Photo by Bernd 📷 Dittrich (https://unsplash.com/@hdbernd?utmsource=karaf&utmmedium=referral) on Unsplash (https://unsplash.com/?utmsource=karaf&utmmedium=referral)

1

Introduction to Data Mining

Definition, Context, and KDD Process

---

Photo by Logan Voss on Unsplash

• Extracting patterns from massive datasets at intersection of machine learning, statistics, and databases
• Interdisciplinary subfield of computer science and statistics
• Analysis step in Knowledge Discovery in Databases (KDD) process
• Involves data pre-processing, model inference, interestingness metrics, visualization
• Misnomer: goal is knowledge extraction, not mining data itself

> Data mining is the process of extracting and finding patterns in massive data sets involving methods at the intersection of machine learning, statistics, and database systems.

— Wikipedia: Data mining

2

Key Data Mining Tasks

Extracting Patterns like Clusters, Anomalies, and Associations

---

Photo by Deng Xiang on Unsplash

• Cluster analysis: groups of similar data records
• Anomaly detection: identifying unusual records
• Association rule mining: discovering dependencies
• Sequential pattern mining: patterns in sequences
• Often uses database techniques like spatial indices

• Patterns summarize large input data
• Enables machine learning and predictive analytics
• Example: identifying groups for accurate predictions by decision systems

3

CRISP-DM Process Model

Most Widely-Used Analytics Model

---

Photo by Deng Xiang on Unsplash

4

Other Methodologies

Beyond CRISP-DM and Key Distinctions

---

Photo by Deng Xiang on Unsplash

• Data Mining vs Data Analysis: uncovers hidden patterns in large data vs testing hypotheses on datasets
• ASUM-DM (IBM, 2015): refines and extends CRISP-DM for predictive analytics
• Data dredging/fishing: risky use on small samples; better for hypothesis generation

Data mining extracts valuable patterns from massive datasets through tasks like clustering, anomaly detection, and association rules. CRISP-DM remains the de facto standard process model.

Leverage these approaches for intelligent insights and predictive power

---

Photo by Deng Xiang on Unsplash