Components of AI Project Framework

False

Reason — Training data is the initial dataset used to train an AI model. Testing data is used to evaluate the performance of the AI model.

False

Reason — Artificial Intelligence is the broader field. Machine learning and deep learning are subcategories of Artificial Intelligence.

True

Reason — The Problem Statement Template aids in summarising all the key points in the 4Ws problem canvas into a single template, which enables us to clearly understand and remember the important aspects of the problem to be solved.

False

Reason — In the AI project cycle, data acquisition is performed first to collect the relevant data, and data exploration is done after acquiring the data to analyse patterns, trends, and relationships.

True

Reason — Testing data is used to evaluate the performance of the AI model. It is the data that the AI algorithm has not seen before and helps to check the accuracy of the AI model.

True

Reason — Tableau is a popular data visualisation tool. It is a user-friendly and interactive tool that supports large datasets and provides options to collaborate with others and securely share data from a variety of sources.

True

Reason — Data can be in the form of text, video, images, audio, and so on.

True

Reason — In the evaluation stage, the AI model is tested using testing data and the output is compared with the expected outcome to assess its accuracy and reliability.

True

Reason — In a system map, if the arrow goes from one element to another with a positive (+) sign, it means that both elements are directly related to each other. If one element increases, the other also increases, and vice versa.

True

Reason — Deep learning models understand how the human brain works in different situations and then try to recreate that behaviour. For example, a deep learning model in a driverless car can identify a person crossing the road.

Identifying a problem and defining its goal

Reason — Problem scoping is the stage where we set clear goals and outline the objectives of the AI project by understanding and defining the problem to be solved.

Initiating

Reason — The AI project cycle consists of five stages: problem scoping, data acquisition, data exploration, modelling, and evaluation. “Initiating” is not included as a stage of the AI project cycle.

Who? What? Where? Why?

Reason — The 4Ws Problem Canvas is a structured framework that helps in problem scoping and consists of four parameters: Who? What? Where? Why?

Open-sourced government portals

Reason — One of the most reliable and authentic sources of information is the open-sourced websites hosted by the government, which provide data in a suitable format that can be downloaded and used.

All of these

Reason — Surveys, web scraping, and sensors and cameras are all listed as suitable sources for data acquisition in an AI project.

Data should be authentic, reliable, and accurate

Reason — Data forms the basis of the AI project, hence, the data has to be authentic, reliable, and accurate. The methods of obtaining data should also be ethical and lawful, and accessing private data without permission is an offence.

Matplotlib

Reason — Matplotlib is a Python library used for creating static, animated, and interactive visualisations in Python.

All of these

Reason — Data visualisation simplifies complex data, helps uncover hidden relationships or anomalies, and provides insights into trends, relationships, and patterns present within the data.

Pie chart

Reason — A pie chart represents data in a circular form divided into sectors, where the size of each sector is proportional to the quantity it represents, and is used to show the proportion or percentage distribution of different categories within a whole.

Machine learning

Reason — Machine learning enables machines to learn on their own and improve with time through experience by learning from data and improving their performance in subsequent iterations.

Deep learning

Reason — Deep learning is based on neural networks with multiple layers that mimic the working of the human brain.

Evaluation

Reason — The evaluation stage focuses on testing the selected AI model using testing data and comparing the results with the expected outcome to assess the accuracy and reliability of the model.

Rule-based approach

Reason — In the rule-based approach, the developer feeds data along with rules to the model, and the machine follows the rules defined by the developer to produce the required output.

Prediction

Reason — Prediction is the process of making forecasts or inferences about unknown or future data using a trained AI model and the patterns it has learned from training data.

System map

Reason — System maps are visual diagrams that help to see and understand the different parts or elements of the AI project and show how all the elements are connected or related to each other.

1. Evaluation is the process of assessing the reliability and efficiency of any AI model.
2. The 4Ws Problem Canvas provides structured framework in the Problem Scoping stage of the AI project cycle.
3. Once a model has been developed and trained, it is important to thoroughly test its Performance and calculate its efficiency.
4. AI Modelling refers to developing a program or algorithm that can be used to draw conclusions or generate predictions.
5. A/An Scatter chart is a graph that shows the relationship between two set of values.
6. Training data is the initial dataset used to train an AI model.
7. Data features describe the type of information that will be collected in response to the Problem statement.
8. The System map helps to find the relationship between the elements of the scoped problem.
9. In the 4Ws canvas, the Who block helps identify the people who are directly or indirectly affected by the problem.
10. In Supervised learning, the model is trained on a labelled dataset.

AI project deployment is the process of implementing an AI model in a real-world scenario. In this stage, the AI model is integrated into the desired software or system and packaged in such a way that it can be used for practical applications. The main goal of AI project deployment is to make the AI model useful for solving real-world problems.

Regression is a type of supervised learning model used to predict continuous numerical values. Continuous data means data that can have any value within a certain range, for example, the price of a product or the salary of an employee. It is used to predict the behaviour of one variable depending on the value of another variable.

Example: A real-life example of regression is predicting the temperature for the next day based on historical weather data, including factors like humidity and cloud cover.

System maps are visual diagrams that help to see and understand the different parts or elements of an AI project. They show how all the elements are connected or related to each other and help in understanding the system’s boundaries and how it interacts with elements in its surroundings.

Any five factors that affect the problem in an AI project are:

1. Stakeholders (Who?) – The people/groups who are directly affected by the problem.
2. Nature of the problem (What?) – What exactly is the issue that needs to be solved (clear problem statement).
3. Evidence that it is a real problem (What?) – Observations/records/reports showing the problem actually exists.
4. Context / Location (Where?) – Where the problem occurs (place, situation, platform, community, etc.).
5. Reason/Impact (Why?) – Why it is worth solving and how it will benefit the stakeholders/society (better quality of life, safety, efficiency, etc.).

Application of Machine Learning — Predicting the weather forecast for the next few days based on historical weather data.

Application of Deep Learning — A driverless car identifying a person crossing the road using image recognition.

Sensors detect changes in the surroundings (environment) such as temperature, pressure, light, sound, distance or motion and convert them into electrical signals or digital data. This data is then processed and analysed by AI systems to take suitable action or decisions.

The AI project cycle consists of five stages. It is a structured framework that outlines the step-by-step process of developing and implementing an AI system. The stages are as follows:

1. Problem Scoping: In this stage, we clearly identify the problem, set clear goals, and define what we want the AI system to achieve. It involves understanding the needs, objectives, constraints, and planning how to solve the problem.
2. Data Acquisition: Here we collect relevant data needed for the AI system from reliable sources (e.g., surveys, sensors, cameras, observations, APIs, web scraping). The data must be accurate and suitable for the problem.
3. Data Exploration: In this stage, we explore and study the data to find patterns, trends, relationships, and possible errors. Data is often represented using charts/graphs to understand it better and decide what information (features) is important.
4. Modelling: After understanding the data, we choose and build an AI model (algorithm/program) and train it using data so that it can make predictions or decisions.
5. Evaluation: Finally, we test the model’s performance by comparing its results with expected outcomes. If the accuracy is good, the model can be used/deployed otherwise, it is improved and tested again.

The two types of data used in the AI project cycle are:

1. Training Data: Training data is the initial dataset used to train an AI model. It is a set of examples that helps the AI model learn and identify patterns or perform particular tasks. The data used for training should be aligned with the problem statement and must be sufficient, relevant, accurate, and wide-ranging so that the model can learn effectively.

2. Testing Data: Testing data is used to evaluate the performance of the AI model. It is the data that the AI algorithm has not seen before. Testing data helps in checking the accuracy and reliability of the AI model and represents the data that the model will encounter in real-world situations.

The following are five data features that can be collected to build an AI system to classify images of animals into different species:

1. Colour (dominant body colours, colour distribution).
2. Body shape and size (long neck, round body, size of head, legs, tail, wings, etc.).
3. Texture (fur, feathers, scales, smooth vs rough appearance).
4. Facial features (eyes, ears, beak, or snout).
5. Patterns/markings (stripes, spots, patches, rings on tail, etc.).

Differences between rule-based approach and learning based approach of AI modelling:

Data exploration is the process of exploring and analysing the collected data to interpret patterns, trends, and relationships. Since data is usually in large quantities, different data visualisation techniques are used to easily understand the data and gain meaningful insights.

Pie Chart

A pie chart represents data in a circular form divided into sectors, where the size of each sector is proportional to the quantity it represents. It is used to understand the contribution of individual categories to the whole and is suitable for showing proportional or percentage distribution of data.

Scatter Chart

A scatter chart is a plot represented with the help of dots, where each dot represents a single data point. It is used to show the relationship between two sets of values and is useful for identifying patterns, correlations, and relationships between two continuous variables.

Applying the 4Ws Problem Canvas to the given scenario:

Scenario:

Developing a mobile app to help users track their daily water consumption

Who? (Stakeholders)

• Users of the mobile app
• Health-conscious individuals
• Families and households
• Healthcare professionals

What? (Nature of the problem)

• Many people do not track their daily water intake.
• Lack of awareness about adequate water consumption can lead to dehydration and health issues.

Where? (Context or location)

• The problem is experienced in daily life, at homes, workplaces, schools, and other places where people carry out their routine activities.
• The solution will be used on mobile devices.

Why? (Benefits of solving the problem)

• Helps users monitor and improve their daily water intake.
• Promotes better health and hydration habits.
• Increases awareness about the importance of drinking sufficient water.
• Contributes to a healthier lifestyle.

For a movie recommendation AI, the data features should describe both the viewer and the movies, so the model can find patterns and recommend accurately.

Important data features for this AI project

1. Viewer features

• Movies/genres the viewer has watched
• Viewer’s ratings/likes/dislikes
• Watch time / completion rate (whether they finished or dropped a movie)
• Search history and movies added to watchlist
• Viewer preferences like language, age group, and preferred content type (action, comedy, family, etc.)

2. Movie (Item) features

• Genre (action, romance, thriller, etc.)
• Language
• Cast and director
• Release year
• Movie duration
• Keywords/tags (e.g., "school life", "sports", "superhero")
• Age rating (U/A, A, etc.)

3. Context features

• Time of viewing (weekend vs weekday, night vs day)
• Device/platform used (mobile/TV)
• What’s trending/popular in the region

These features help the AI learn relationships between viewer choices and movie characteristics, and then recommend movies that match the viewer's taste.