Class 12 | AI Notes | Subject Code 843
ailogicschool.blogspot.com · CBSE 2025–26
Class XII Artificial Intelligence
Chapter-wise Notes
Subject Code 843 · Job Role: AI Assistant · Total Marks: 100 (Theory 50 + Practical 50)
Part A: Employability Skills — 10 Marks
Part B: Subject Specific Skills — 40 Marks
Part C: Practical Work — 50 Marks
📊 Unit-wise Marks Distribution
| Part | Unit | Hours | Theory | Practical |
|---|---|---|---|---|
| PART A Employability |
Communication Skills-IV | 15 | 2 | — |
| Self-Management Skills-IV | 10 | 2 | — | |
| ICT Skills-IV | 15 | 2 | — | |
| Entrepreneurial Skills-IV | 10 | 2 | — | |
| Green Skills-IV | 10 | 2 | — | |
| Part A Total | 60 | 10 | — | |
| PART B Subject Skills |
U1 Python Programming – II* | 6+18 | —* | Practical only |
| U2 Data Science Methodology | 8+12 | 8 | 12 | |
| U3 Making Machines See | 6+12 | 6 | 12 | |
| U4 AI with Orange Tool* | 4+18 | —* | Practical only | |
| U5 Big Data and Analytics | 7+12 | 6 | 12 | |
| U6 Understanding Neural Networks | 8+12 | 8 | 12 | |
| U7 Generative AI | 6+12 | 7 | 12 | |
| U8 Data Storytelling | 5+4 | 5 | 4 | |
| Part B Total | — | 40 | 50 | |
| PART C Practical |
Capstone Project + Practical File + Lab Test + Viva | — | — | 50 |
| GRAND TOTAL | — | 50 | 50 | |
🗣️ Communication Skills-IV
Advanced communication, professional writing, presentation skills, public speaking
15 hrs · 2 marks
🧘 Self-Management Skills-IV
Stress management, goal setting, time management, emotional intelligence
10 hrs · 2 marks
💻 ICT Skills-IV
Advanced ICT, digital tools, cybersecurity awareness, cloud services
15 hrs · 2 marks
🚀 Entrepreneurial Skills-IV
Business planning, innovation, startup ecosystem, financial literacy
10 hrs · 2 marks
🌿 Green Skills-IV
Sustainability, environment conservation, green technology, SDG alignment
10 hrs · 2 marks
📌 Note: Detailed curriculum for Part A: Employability Skills can be downloaded from the official CBSE website. These units are NOT included in the Subject Specific skills evaluation but carry 10 marks in theory.
NumPy
Pandas
CSV Files
Missing Values
Linear Regression ★ Advanced
📚 Sub-topics
1.1Recap of NumPy Library — arrays, indexing, slicing, mathematical operations
1.2Recap of Pandas Library — Series, DataFrame creation, basic operations
1.3Importing & Exporting Data between CSV Files and DataFrames
1.4Handling Missing Values — isnull(), fillna(), dropna() strategies
1.5★Linear Regression Algorithm — For Advanced Learners only (not in exam)
🎯 Learning Outcomes
✦Apply fundamental concepts of NumPy and Pandas libraries to perform data manipulation and analysis tasks
✦Import and export data between CSV files and Pandas DataFrames, ensuring data integrity and consistency
✦Handle missing values in a dataset using appropriate strategies (mean/median/mode imputation)
✦Implement Linear Regression algorithm on Google Colab or any Python IDE (Advanced Learners)
🔬 Practicals / Activities
P1
Write Python code to create a Pandas DataFrame using any sequence data type. Display full DataFrame, first 5 records, last 10 records, and count missing values.
P2
Download a CSV dataset from a public open-source website. Read it into a Pandas DataFrame, perform statistical functions, check and fill missing data.
P3
Import and Export Data between CSV Files and DataFrames (round-trip verification).
P4★
Implement Linear Regression algorithm — train/test split demonstration using Python. Advanced Learners
Key Functions to Know
pd.DataFrame()
df.head() / tail()
df.isnull().sum()
df.fillna()
df.dropna()
pd.read_csv()
df.to_csv()
df.describe()
np.reshape()
np.linspace()
train_test_split()
LinearRegression()
* Note: Unit 1 is evaluated ONLY in the Practical examination (Lab Test). It does NOT carry separate marks in the Theory paper.
MSE
RMSE
Confusion Matrix
Precision
Recall
F1 Score
Capstone Project
📚 Sub-topics
2.1Introduction to Data Science Methodology
2.2Steps for Data Science Methodology — Problem → Data → Model → Deployment
2.3Model Validation Techniques — Train-Test Split, Cross-Validation, k-Fold
2.4Model Performance Evaluation Metrics — MSE, RMSE, Accuracy, Precision, Recall, F1
2.5Confusion Matrix — TP, TN, FP, FN interpretation
2.6Integration with Capstone Project
📐 Important Formulas
MSE = Σ(actual−pred)²/n
RMSE = √MSE
Accuracy = (TP+TN)/Total
Precision = TP/(TP+FP)
Recall = TP/(TP+FN)
F1 = 2×(P×R)/(P+R)
🎯 Learning Outcomes
✦Integrate Data Science Methodology steps into the Capstone Project
✦Identify the best way to represent a solution to a problem
✦Understand the importance of validating machine learning models
✦Use key evaluation metrics for various machine learning tasks
🔬 Practicals / Activities
P1
Calculate MSE and RMSE values for given data using MS Excel.
P2
Calculate Precision, Recall, F1 Score, and Accuracy from a given Confusion Matrix.
P3
Python Code to Evaluate a Model — compare predicted vs actual, compute all metrics. (To be evaluated in practicals only)
Computer Vision
Binary Art
Image Processing
OpenCV ★ Advanced
ML Model on Web
📚 Sub-topics
3.1How Machines See — pixels, images as matrices of numbers (0–255)
3.2Working of Computer Vision — the 4-stage pipeline
3.3Computer Vision Process — Acquisition → Preprocessing → Feature Extraction → Analysis
3.4Applications of Computer Vision — healthcare, surveillance, autonomous vehicles, agriculture
3.5Challenges — ethics, privacy, bias, computational cost, technical limitations
3.6The Future of Computer Vision — trends and transformative potential
🎯 Learning Outcomes
✦Explain Computer Vision and its significance in visual data analysis
✦Understand key stages: acquisition, preprocessing, feature extraction, and analysis
✦Identify real-world applications in healthcare, surveillance, and autonomous vehicles
✦Analyze challenges such as ethics, privacy, and technical limitations
✦Develop basic skills in using OpenCV and deploying ML models online
🔬 Practicals / Activities
P1
Binary Art — Recreating images with 0s and 1s using NumPy and matplotlib. Demonstrate how machines represent visual data numerically.
P2
Creating a Website Containing an ML Model — deploy a simple image classification model online.
P3★
Working with OpenCV — load, display, resize images; apply grayscale conversion, Gaussian blur, and Canny edge detection. Advanced Learners
Key Concepts
Pixel (0–255)
Grayscale image
RGB image (3-channel)
Thresholding
Edge detection
cv2.imread()
cv2.cvtColor()
cv2.Canny()
cv2.resize()
Feature extraction
Data Mining
Widgets
Classification
NLP
Computer Vision
Word Cloud
📚 Sub-topics
4.1What is Data Mining? — definition, process, goals
4.2Introduction to Orange Data Mining Tool — interface overview
4.3Beneficiaries of Orange Data Mining — who uses it and why
4.4Getting Started with Orange Tool — installation, canvas, widgets
4.5Components of Orange — widgets, links, canvas, workflow
4.6Default Widget Catalogue — Data, Visualize, Model, Evaluate, Unsupervised
4.7Key Domains — Data Science, Computer Vision, NLP with Orange
🔬 Practicals / Activities
P1
Load and visualize the Iris dataset using Scatter Plot, Data Table, and other visualization widgets.
P2
Use Classification widgets (kNN, Decision Tree, Naive Bayes) to classify the Iris dataset.
P3
Evaluate the Classification Model using Test and Score + Confusion Matrix widgets. Read AUC, CA, F1, Precision, Recall.
P4
Computer Vision with Orange — image analytics using Image Embedder and Distances widgets.
P5
Natural Language Processing with Orange — load corpus, preprocess text, create Word Cloud to visualize word frequencies.
Key Orange Widgets to Know
Datasets
Data Table
Scatter Plot
Box Plot
Distributions
kNN
Decision Tree
Test and Score
Confusion Matrix
Preprocess Text
Word Cloud
Bag of Words
📌 Practical File: Take screenshots of every step and widget in Orange and paste them in your practical file. A minimum of 3 Orange programs must be included in the Practical File.
5 V's of Big Data
Data Analytics
Data Streams
Orange Analytics
📚 Sub-topics
5.1Introduction to Big Data — definition, history, why it matters
5.2Types of Big Data — Structured, Semi-structured, Unstructured
5.3Advantages and Disadvantages of Big Data
5.4Characteristics — 5 V's: Volume, Velocity, Variety, Veracity, Value
5.5Big Data Analytics — descriptive, diagnostic, predictive, prescriptive
5.6Working on Big Data Analytics — tools and workflow
5.7Mining Data Streams — real-time processing concepts
5.8Future of Big Data Analytics — trends and AI integration
The 5 V's of Big Data — Must Know for Exams
Volume — Size of data (petabytes)
Velocity — Speed of data generation
Variety — Types (text, image, video)
Veracity — Accuracy / trustworthiness
Value — Usefulness of the data
🔬 Practicals / Activities
P1
Performing Big Data Analytics with Orange Data Mining tool — load a large dataset, use grouping, filtering, and visualization widgets. (Evaluated in practicals only)
P2
Python-based Big Data Analysis — use Pandas groupby(), value_counts(), pivot_table() on a CSV dataset with 500+ rows. Generate charts.
Neurons
Layers
Activation Functions
TensorFlow
Keras ★ Advanced
Backpropagation
📚 Sub-topics
6.1Parts of a Neural Network — Input layer, Hidden layers, Output layer
6.2Components — Neurons, Weights, Biases, Activation Functions
6.3Working of a Neural Network — forward propagation, loss, backpropagation, gradient descent
6.4Types of Neural Networks — ANN, CNN, RNN, LSTM, GAN, Transformer
6.5Future of Neural Networks and Societal Impact
Activation Functions — Must Know
ReLU — hidden layers
Sigmoid — binary output
Softmax — multi-class
Tanh — hidden layers
Linear — regression
🔬 Practicals / Activities
P1
Explore Machine Learning for Kids — create a neural network for identifying animals and birds (no-code platform).
P2
Build a TensorFlow model to convert Celsius to Fahrenheit. Observe how it learns F = 1.8C + 32. (Evaluated in practicals only)
P3
Classification problem using TensorFlow Playground — adjust layers, neurons, activation function at playground.tensorflow.org. Take screenshots.
P4★
Use Python Keras to create and train a multi-layer neural network predicting Fahrenheit from Celsius (or Iris classification). Advanced Learners
LLM
Generative vs Discriminative
Gemini API
ChatGPT
Canva AI
Ethics
📚 Sub-topics
7.1Introduction to Generative AI — what it is, history, examples
7.2Working of Generative AI — training on data, generating new samples
7.3Generative vs Discriminative Models — create new vs classify existing
7.4Applications — text, image, audio, video generation; code writing
7.5LLM — Large Language Model — GPT, Gemini, Claude, LLaMA
7.6Future of Generative AI — AGI debate, multimodal models
7.7Ethical and Social Implications — deepfakes, copyright, bias, misinformation
Key Differences — Generative vs Discriminative
Generative: Creates new data
Discriminative: Classifies data
Gen. Examples: GPT, DALL-E, Sora
Disc. Examples: Spam Filter, SVM
Gen. learns P(X)
Disc. learns P(Y|X)
🔬 Practicals / Activities
P1
Signing up for Canva — explore AI image generation features (Text to Image, Magic Design).
P2
Animaker AI Video Generation — create a short educational video using AI-powered tools.
P3
Use Google Gemini to craft prompts and generate text outputs. Practice prompt engineering techniques.
P4
Explore ChatGPT for conversational text generation and creative tasks — compare outputs with different prompts.
P5★
Write Python code to initialize the Gemini API and create a multi-turn chatbot. Advanced Learners
Narrative
Freytag's Pyramid
Visualization
MDMS Case Study
Ethics
📚 Sub-topics
8.1Introduction to Storytelling — why stories work, elements of a good story
8.2Elements of a Story — character, conflict, resolution, setting
8.3Introduction to Data Storytelling — combining data + visuals + narrative
8.4Why is Data Storytelling Powerful? — impact on decision making
8.5Essential Elements of Data Storytelling — data accuracy, clear visuals, audience-aware narrative
8.6Freytag's Pyramid — Exposition → Rising Action → Climax → Falling Action → Resolution
8.7Types of Data and Visualizations — bar, line, pie, scatter, heat map — when to use each
8.8Steps to Create a Story Through Data — 6-step process
8.9Ethics in Data Storytelling — misleading charts, cherry-picking, correlation vs causation
Freytag's Pyramid Applied to Data Stories
1. Exposition — Context & background data
2. Rising Action — Trend building
3. Climax — Key insight / turning point
4. Falling Action — Analysis, causes
5. Resolution — Conclusion & call to action
🔬 Practicals / Activities
P1
Create an effective data story using given data — follow all 5 stages of Freytag's Pyramid.
P2
MDMS Case Study (Mandatory): Using student enrollment, attendance, and dropout data, create a data story about the impact of Mid-Day Meal Scheme since 1995. Include visualizations, charts, and analysis of external factors. Provide comprehensive narrative highlighting MDMS vs dropout rate relationship.
📌 Practical File: Minimum 1 Data Story must be included in the Practical File. The MDMS story is the recommended sample from CBSE curriculum.
📁 Practical File Requirements
🐍 Python Programs (min. 6)
🍊 Orange Programs (min. 3)
📖 Data Story (min. 1)
📌 Orange Snapshots: All steps and outputs in Orange must be screenshotted and pasted in the Practical File. Similarly for TensorFlow Playground and optional programs.
🚀 Capstone Project Guidelines
Group size: minimum 3, maximum 5 students per group
Project must be aligned with any of the 17 UN Sustainable Development Goals (SDGs)
Complete the project in Class XII AND submit full project documentation
Video must be exactly 3 minutes — include: Problem Statement, SDG alignment, AI concept/algorithm used, working demo, conclusion, acknowledgement
Use Python OR no-code/low-code platforms like Orange Data Mining for development
Follow CBSE-IBM AI Project Cookbook guidelines (Page 2) for documentation format
💻 Required Hardware & Software
HWCore i5 or equivalent, min 8GB RAM, 512GB SSD, 17" LED Monitor, Webcam, Mic, 100Mbps WiFi
OSAny OS with antivirus activated
PYPython IDLE + Anaconda Navigator (NumPy, Pandas, Matplotlib, Scikit-learn)
OROrange Data Mining Tool (latest version)
SWGoogle Suite / Productivity Suite for documentation
Comments
Post a Comment