Kips Class 9 AI Unit 2: AI Project Cycle

Unit 2: AI Project Cycle

Question 1: Explain the various stages of AI project cycle.
Answer: Planning → Data → Modelling → Evaluation and Deployment
Stage 1: Planning: It is indispensable for the success of the projects, and AI projects are no exception. Goals of AI can be divided into two categories
• Analysing things which humans can do, but not very well if the data is large.
• Making predictions or forecasts which human beings are not good at.
Problem Scoping
1. Business or Project requirements: This is the first step in an AI project cycle. It starts with determining which involve asking questions.
2. Analytics approach to the Project: The second step in the AI project cycle involves finding out how to achieve the project requirements, objects or goals determined in the first step.

Stage 2: Data: AI systems are dependent on large amount of data. Without this data, we cannot do anything at all.
Data Acquisition
3. Data Requirements: It deals with identifying the data requirements of the project.
4. Data Collection: It deals with finding out how to collect this data. Links have to be established with people and sources, and methods need to be found for extracting the required data from these sources. Example: survey, feedbacks, website etc.
Data Exploring
5. Exploratory Data Analytics: It deals with validating or verification of the collected data to see if it meets the requirements or not.
6. Data Preparation: It involves preparation or curation of the data.

Stage 3: Modelling: Data modelling refers to selecting the specific algorithms and building AI models for working with the data.
7. Model Selection: Modelling involves selection of some forms of AI data model for solving the problems and achieving goals. The selection of the model basically depends on the problem and the data being used. Example: Neural Networks, Decision Trees are some models.

Stage 4: Evaluation and Deployment: It involves testing of the system, and the system deployment.
8. Project Evaluation: It involves testing the system. This stage will also involve making modifications to the project for ensuring that it functions as it was originally intended to function.
9. Project Deployment: It involves the deployment of the project. The projects have to deployed for making them affective. An un-deployed project is completely useless. Project deployment is done after the project clears the testing and evaluation stage.
10. System Tuning: It involves re-training and re-tuning of AI system at regular intervals for ensuring their accuracy and effectiveness.
11. Project Review: It helps in separating the code for future use and finding out the areas for training and development of the team.

Question 2: Define Big Data.
Answer: It refers to the data that cannot be practically fit into the standard relational databases like Oracle, SQL Server, MS Access etc. The size of this data is so large that the traditional databases cannot be used for capturing, managing and processing this data.

Question 3: Explain evaluation and deployment stage.
Answer:
Evaluation and Deployment: It is the fourth stage in AI project cycle. It involves testing of the system, and the system deployment.
8. Project Evaluation: It involves testing the system. This stage will also involve making modifications to the project for ensuring that it functions as it was originally intended to function.
9. Project Deployment: It involves the deployment of the project. The projects have to deployed for making them affective. An un-deployed project is completely useless. Project deployment is done after the project clears the testing and evaluation stage.
10. System Tuning: It involves re-training and re-tuning of AI system at regular intervals for ensuring their accuracy and effectiveness.
11. Project Review: It helps in separating the code for future use and finding out the areas for training and development of the team.

Question 4: Mention 5 Questions that help in identification of the AI problem for creating AI projects.
Answer: 5 Questions that help in identification of the AI problem for creating AI projects are:
• How can we make better decisions?
• Where are we most inefficient?
• Where do we have a lot of relevant data?
• What goals do we want to achieve?
• Will this actually solve our problems?