Unit 5: Data Literacy – Collection to Analysis4 Levels of Measurement, Statistics, Matplotlib & Matrices in AI
Master the data foundation of AI. Learn primary vs secondary data collection, 4 levels of measurement (Nominal, Ordinal, Interval, Ratio), central tendency and dispersion calculations, Matplotlib plotting, matrix operations for computer vision, and preprocessing pipelines.
1. What is Data Literacy & Data Collection Sources
Data is the fuel powering artificial intelligence. Over 90% of the world's data was created in just the last two years. Data formats fall into three categories:
Organized into strict tabular rows and columns (e.g. relational SQL databases, Excel, CSV files with student records).
Contains tags and metadata identifiers without rigid table schemas (e.g. JSON, XML, social media posts with hashtags).
Lacks any predefined organizational structure (e.g. digital photos, audio recordings, customer reviews, video streams).
Primary vs. Secondary Data Collection
Primary Sources (Firsthand)
- • Surveys & Questionnaires: Sampling populations for consumer feedback.
- • Interviews: Direct structured/unstructured dialogues.
- • Observations: Recording events as they unfold.
- • Controlled Experiments: Manipulating variables to establish cause-effect.
Secondary Sources (Pre-Existing)
- • Web Scraping: Automated tools extracting HTML content and prices.
- • Social Media Tracking: Public tweets, comments, sentiment trends.
- • Satellite Data: Earth observation weather & agriculture imagery.
- • Open Data Repositories: Kaggle, GitHub, government databases.
2. The Four Levels of Measurement
Nominal Scale ("In Name Only")
Categories or labels with no natural order or numeric ranking. Mathematical calculations cannot be performed.
Examples: Eye colors, gender, car manufacturers (BMW, Audi), sports jersey numbers.
Ordinal Scale (Ordered Ranks)
Categories follow a strict order or ranking, but the exact differences between ranks cannot be mathematically quantified.
Examples: Meal rating (unpalatable → delicious), letter grades (A, B, C, D), competition standings.
Interval Scale (Arbitrary Zero)
Ordered numerical data with measurable, uniform intervals between values, but no absolute true zero point. Differences make sense, but ratios do not.
Examples: Temperature in °C and °F (80°C is not 4x as hot as 20°C; 0°C is not absence of heat).
Ratio Scale (Absolute True Zero)
Features consistent measurable intervals AND a true natural zero point representing complete absence. Multiplication and division ratios are fully valid.
Examples: Weight in kg, height in cm, bank account balance, exam marks out of 100 (80 marks is 4x 20 marks).
3. Central Tendency & Dispersion with Python
Mean = Σfx / n. Ideal for evenly spread data with no extreme outliers.
Middle value when data is sorted. Highly robust against extreme outliers and skewed distributions.
Most frequently occurring value. Used in inventory management (e.g. printing the most popular books).
Heights at shoulder: 600mm, 470mm, 170mm, 430mm, 300mm
(600+470+170+430+300)/5 = 394 mm
[206² + 76² + (-224)² + 36² + (-94)²]/5 = 21,704 mm²
√21704 = 147.32 mm
4. Matrices and Their Role in Computer Vision
A Matrix is a rectangular arrangement of numbers into $m$ rows and $n$ columns (order $m \times n$). In Artificial Intelligence, matrices are ubiquitous:
A digital image is represented as a matrix of numerical pixel intensities (0 to 255 in 8-bit grayscale; 3 stacked matrices for RGB).
User-item interaction matrices where rows represent users and columns represent product ratings.
High-dimensional word embedding vectors are 1-dimensional matrices encoding semantic relationships.
Official Handbook MCQs & Answer Key
Click any option to instantly see if you're correct with the official CBSE explanation.
Which of the following best defines 'Data Literacy' in the context of Artificial Intelligence?
What is the primary purpose of data preprocessing before training machine learning models?
How can missing values (NaN) be handled in a dataset during preprocessing?
Which of the following statements about dataset quantity and quality in machine learning is true?
Which of the following is an example of a PRIMARY source of data collection?
What primary data collection method involves direct communication with individuals or focus groups?
Which of the following measurements is an example of RATIO scale data?
What is the key distinguishing feature that separates Ratio scale data from Interval scale data?
Which statistical measure of central tendency is most suitable for symmetrical data with no extreme outliers?
What is the term used to describe the graphical or pictorial representation of statistical data using charts and plots?
Unit 5 Quick Recall Cheat Sheet
Frequently Asked Questions (FAQ)
Master Data Literacy & Statistics with 1:1 Live Mentorship
Master statistical analysis, Python data visualization, and matrix operations with personalized 1-on-1 tutoring from certified data scientists. Ace your Class 11 AI board exams and lab practicals.