AI Project Cycle Notes

This document outlines the AI project cycle, a structured approach designed to facilitate the completion of AI project
s. The cycle is composed of six distinct stages: Problem Scoping, Data Acquisition, Data Exploration, Modeling, Eval
uation, and Deployment. Each stage plays a crucial role in ensuring that AI projects are executed efficiently and
effectively, ultimately leading to better outcomes. The notes also highlight the importance of ethical consideration
s throughout the cycle, including data privacy and bias mitigation.

Evaluation Deployment

Validate Models Monitor Models

Test Performance Update Models

Problem Scoping Data Acquisition

Define Objectives Collect Data

Identify Stakeholders Ensure Quality
AI Project
Modeling Cycle Data Exploration

Select Algorithms Analyze Patterns

Train Models Visualize Data

Ethical Considerations

Data Privacy
Bias Mitigation

1. Problem Scoping
Definition: Problem Scoping is the first step in the AI Project Cycle. It involves
identifying and defining the problem you want to solve using AI. This step is crucial
as it sets the direction for the entire project.
• Context Understanding: It requires a deep understanding of the problem's context, id
entifying stakeholders, and defining the project's goals.
• 4Ws Problem Canvas: This tool helps to identify key elements related to the problem:
• Who: Stakeholders
• What: The problem and evidence
• Where: Context and location
• Why: Benefits of solving the problem
• Problem Statement Template: This summarizes the key points derived from the 4Ws
Canvas.

4Ws Problem Canvas

Definition
Who
What Identify Problem
Where Define Problem
Why Problem
Scoping Context Understanding
Problem Statement Template
Stakeholders
Summarize Key Points Project Goals
Derived from 4Ws

2. Data Acquisition
• Definition: Data Acquisition involves gathering the necessary data for the AI project.
• Data Features: These are determined by the problem statement and dictate the types
of data that need to be collected.
• Source Reliability: It is crucial to acquire data from reliable and authentic sources, such
as open-sourced government websites, to ensure the project's efficiency.
• System Maps: These visual tools help to illustrate the relationships between different
elements of the problem, aiding in strategizing solutions.

Definition Data Features

Gathering Data Problem Statement

AI Project Data Types
Data
System Maps Acquisition Source Reliability

Illustrating Relationships Reliable Sources

Strategizing Solutions Authentic Sources

3. Data Exploration
• Definition: Data Exploration involves analyzing and visualizing the acquired data to ide
ntify trends, patterns, and relationships. This step also helps in identifying any
anomalies or missing values and provides insights into how to pre-process the data.
• Visualization Importance: Visualizing data is essential for understanding its trends,
• defining a strategy for selecting the right AI model, and effectively communicating findi
ngs.
• Graph Types: Various types of graphs and visualizations, such as bar graphs, pie charts,
and line graphs, can be utilized for data representation and interpretation.
• Tools: Tools like Power BI, Tableau, Excel can assist in data visualization and analysis.

Data
Exploration

Analyze and
Visualize
Data

Identify
Trends,
Patterns,
Relationships

Importance Graph Types Tools

of
Visualization

Understand Define AI Communicate Bar Graph Pie Chart Line Graph Tableau
Trends Model Findings
Strategy

4. Modeling
• Definition: Modeling is the process of creating an AI model that can learn from the data and
make predictions or decisions. This involves selecting and developing the appropriate AI
algorithms or models to address the defined problem.
• Approaches: There are two main approaches to AI modeling:

• Rule-Based: Follows predefined rules set by developers.

• Learning-Based: Learns from data and adapts its algorithms accordingly.

Data-Driven
Predefined Rules
Learning

Predictable Adaptive
Outcomes Algorithms

Rule-Based Learning-Based

Comparing Rule-Based and Learning-Based AI Modeling

Approaches

5. Evaluation

AI Assistance Readiness

AI's Willingness

AI's Readiness

Core Message

• Definition: Evaluation involves assessing the performance and effectiveness of the

trained AI models using testing data.
• Importance: This stage is crucial for identifying areas for improvement and ensuring
the model's accuracy and reliability.
• Key Metrics: Important metrics and terminologies used in model evaluation include
accuracy, precision, recall, F1-score, and ROC curves.

6. Deployment
• Definition: Deployment is the final stage where the developed AI model or solution is
implemented in a real-world setting.
• Integration: This involves integrating the model into existing systems or deploying it
as a standalone application.
• Planning and Execution: Careful planning and execution are required to ensure a
smooth transition and successful integration.

Deploy an AI Model in a Real-World Setting

Integrate the Deploy the AI

AI model into Plan and model in a
existing execute the real-world
systems deployment setting

Throughout the AI project cycle, ethical considerations are paramount. This includes
addressing data privacy, mitigating bias in data and algorithms, and promoting responsible
AI development.