Lecture on AI, Digitalization, and Data
Introduction
- Discussion during the initial waiting period
- Digital Twin: Understanding and real-time replication
- Clarification of concepts and the use of AI in predicting outcomes
Digitalization and Digitization
- Digitization: Converting analog to digital (e.g., scanning documents)
- Digitalization: Using digital data to improve processes, workflows, and business models
- Examples: Expense report automation (digitization) vs. end-to-end process automation (digitalization)
Technologies in Digitalization
- IoT: Collects data (e.g., smartwatches)
- Cloud: Stores data (internal and external clouds, regulatory compliance)
- AI/ML/DS: Processes data (Artificial Intelligence, Machine Learning, Data Science)
- BI/AR/VR: Decision making and visualization (Business Intelligence, Augmented Reality, Virtual Reality)
- RPA: Automates processes (Robotic Process Automation, e.g., chatbots)
- Cyber Security: Secures data
- Blockchain: Verifies and connects systems
Historical and Current Perspectives on AI
- Statistics: Analyze past data (e.g., average height in a class)
- Machine Learning: Predict future outcomes based on training data (e.g., loan approvals)
- Data Science: Combines machine learning with engineering tools for large-scale data analysis
- Deep Learning: Analyzes unstructured data (e.g., images, texts)
Importance of Business Collaboration
- Features and data selection should be guided by domain experts
- Understanding and involvement of domain experts are critical for project success
Data Types
- Nano Data: <5,000 rows
- Medium Data: ~50,000 rows
- Big Data: >100,000 rows
Data Challenges and Methodologies
- Handling missing data and outliers
- Anomaly detection and its importance
- Use of public data for comparative analysis
- Feature selection by understanding the domain
AI Models and Techniques
- Classification Models: Logistic regression, decision trees, K-nearest neighbors
- Anomaly Detection: Identifying outliers and unusual patterns
- Ensemble Models: Combining multiple models for better prediction
- Regression Models: Predicting a value using linear regression
Assignment 1 Overview
- Create a business problem solution with AI integration
- Include elements like Blueprint, Infrastructure, Data Requirements, Policy, etc.
- Guidelines for creating slides/papers
Closing Remarks
- Explanation to continue in the next class with more on unstructured data and Generative AI
Please use the provided guidelines and ensure clarity in responses for the assignment.