CBSE Class 6 AI Chapter 2: Basic Data Concepts — Complete Teaching Guide

Chapter 2 of the Class 6 Artificial Intelligence unit is Basic Data Concepts, allocated 05 hours out of the 20-hour AI block. The single most important thing about teaching it: this is not a chapter about drawing bar graphs. It is the chapter where students first learn that data can be organised honestly or dishonestly, and that a machine can only ever be as sensible as the data it was handed.

Class 6 sits inside a 100-hour Computational Thinking and AI allocation — 40 hours of advanced computational thinking, 20 hours of introductory AI across four chapters of five hours each, and 40 hours of interdisciplinary projects. Chapter 2 is the second of those four. You can see where it sits in the full Class 6 CT and AI syllabus, and how the whole scheme works in the 2026–27 curriculum overview.

There is no coding here. No Python, no spreadsheets required, no AI tool that students must log into. A teacher with chalk, a blackboard and thirty-six students in the room has everything the chapter needs. What the chapter demands instead is a teacher willing to ask an uncomfortable question at the end of every activity: what could someone wrongly conclude from this data?

One sequencing decision matters more than any other, and most schools get it wrong. The NCERT Class 6 mathematics chapter Data Handling and Presentation teaches tally marks, frequency tables, bar graphs and pictographs. Teach that first. This is a recommendation, not a CBSE rule — the handbook does not mandate an order — but if you run Chapter 2 before the maths chapter, you will spend three of your five hours teaching bar-graph mechanics and none of them teaching the thinking. Coordinate with the maths teacher in the first week of the term.

What the syllabus actually says

The CBSE handbook table for Class 6, AI Chapter 2, reads:

Basic Data Concepts — Introduction to data types (numbers, text, images, sound); Simple data organisation and representation using tables or charts — 05 hours

That is the whole entry. Two topics, five hours. Everything below is an unpacking of those two lines, not an addition to them. Where I suggest something the handbook does not require, I say so.

The concepts, explained

What data actually is

Start with a definition students can hold on to: data is a record of something that happened, stored so it can be used later. Not “numbers”. Not “information on a computer”. A record.

The attendance register is data. The mid-day meal count the cook writes on the kitchen wall is data. The rainfall figure the newspaper prints in millimetres is data. A WhatsApp voice note from a student’s mother is data. Your school’s admission forms in a steel almirah are data. None of that requires a computer.

Then draw the distinction that Chapter 1 set up. In Chapter 1, Introduction to AI, students learned that an AI system learns from examples rather than following a fixed rule. Those examples are data. A face-unlock system saw thousands of photographs. A voice assistant heard thousands of recordings. If the school van’s GPS app knows the traffic on the Ring Road at 7:40 a.m., it is because thousands of phones sent their positions.

Say the sentence out loud and put it on the board: no data, no AI. That is the through-line of the chapter and the reason it exists inside an AI unit rather than only inside maths.

The four data types, and the one thing they have in common

The syllabus names four: numbers, text, images, sound. Teach them as four appearances of the same underlying thing.

• Numbers. 27 students present. 68 mm of rain on Tuesday. A score of 148 for 6. A recharge of ₹239. Split this into counts (whole things you can tally) and measurements (things you read off an instrument — temperature, height, rainfall). Students confuse the two constantly, and it matters later because you cannot sensibly average a count of “auto/rickshaw” but you can average a temperature.
• Text. Names, answers, addresses, the message “Bring your ID card tomorrow”. A computer does not store the letter N. It stores a number that has been agreed to mean N. This is the moment students first meet the idea of an encoding, and it is enough at Class 6 to say “there is an agreed code table” without naming ASCII or Unicode.
• Images. A photograph is a grid of tiny squares called pixels. Each square holds numbers describing its brightness or colour. Zoom into any photo on a phone until the squares appear — that thirty-second demonstration does more work than ten minutes of explanation.
• Sound. A microphone measures the air pressure thousands of times a second and writes down each measurement. Play back the list of numbers fast enough and you hear the voice again. A one-minute WhatsApp voice note is, quite literally, a very long list of numbers.

Then the payoff line, which is the whole point of the topic: whatever it looks like on the outside, inside the machine everything is numbers. That is why a single machine can handle a photograph, a song and a mark sheet.

Collecting data: where the errors are born

Everything that goes wrong with a conclusion went wrong earlier, usually at collection. Give students a five-stage pipeline and make them name the failure mode at each stage. This is the vocabulary they will reuse in Class 7, where machine learning is introduced, and again in the Class 8 AI project lifecycle.

Three collection failures are worth drilling because Class 6 students commit all three within the first ten minutes of any survey:

1. Asking the wrong people. A student surveys their four friends on the front bench and reports it as “the class prefers”. Six absentees on the day of the survey means the data describes only who came.
2. Asking a leading question. “You like the mid-day meal, don’t you?” is not the same question as “What did you think of today’s mid-day meal?” One of them manufactures its answer.
3. Allowing categories that overlap or leak. If the options are walk, cycle, bus, car, and a student comes by cycle for half the way and walks the rest, where does the tally mark go? Decide the rule before collecting, and write it down.

Add a habit that costs nothing: record the total number of people asked, the date, and who was absent. A dataset without that context cannot be judged. This connects directly to the abstraction work in the computational thinking half of the year — see our Class 6 abstract thinking module for how deciding what to leave out is itself a skill.

Organising data: tally to frequency table

Raw data is unusable. Thirty-six shouted answers are noise until they are recorded in a structure. That structure — a tally sheet becoming a frequency table — is the first act of organisation students perform, and it maps exactly onto what the NCERT maths chapter Data Handling and Presentation teaches.

Insist on three non-negotiables:

• Tally marks in groups of five, with the fifth stroke laid diagonally across the previous four. Not because it is prettier, but because it makes recounting possible.
• A total row. If the class has 36 students and the tallies add to 41, something is wrong and it must be found before anything is drawn.
• Column headings with units. “Rainfall (mm)” not “Rainfall”. A number without a unit is not data.

The frequency table is the pivot point of the whole chapter. Everything before it is collection; everything after it is presentation. Students who can build a clean frequency table can build any chart the syllabus asks for.

Representing data: tables and charts

The syllabus says “tables or charts”. Teach the choice, not just the drawing. A table is better when exact values matter — a mark sheet, a rainfall record, a fee register. A chart is better when comparison matters — which mode of travel dominates, whether attendance is rising.

For a bar chart, the checklist is short and should be memorised:

1. A title that says what and who. “How Class 6 travels to school”, not “Bar Graph”.
2. Both axes labelled, with units where relevant.
3. A scale that starts at zero and moves in equal steps.
4. Bars of equal width with equal gaps.
5. The value written above each bar. This costs one second and prevents half of all misreadings.

Pie charts are where Class 6 goes wrong, usually because the drawing software makes them easy. A pie chart answers exactly one question: what share of the whole is each part? It fails the moment there are too many slices or the slices are similar in size. Give students a working rule: more than five slices, or two slices you cannot tell apart by eye, means use a bar chart instead.

Reading a chart honestly

This is the section that makes Chapter 2 an AI chapter. Two questions, asked every single time:

Question one: what does this chart show? Students answer this easily. Walking is the most common. Twelve students walk.

Question two: what could someone wrongly conclude from this data? This is harder and far more valuable. From the travel chart, wrong conclusions include: “most Indian children walk to school” (one class, one school, one city); “the school bus is unpopular” (perhaps the bus does not go to those localities at all); “nobody in Class 6 owns a car” (the question was about travel, not ownership); “this is how they travel every day” (it was one day, and it was July — ask again in December).

You are teaching sampling and bias without using either word. The idea underneath is simply: the data describes who you asked, when you asked, and what you asked — nothing more. When students meet training data in Class 7, and biased training data in the Class 8 project lifecycle, this is the hook the new learning attaches to. It also connects forward to Chapter 4 on AI ethics and digital responsibility, where fairness stops being abstract.

Then show them the mechanical trick that makes charts lie: cutting the y-axis. Two charts, identical numbers, opposite impressions.

How to teach it — the five hours

Period 1 should not begin with a definition on the board. Send students out for ten minutes with a clipboard to find data in the school — the notice board, the attendance register, the water-cooler temperature sticker, the fee receipt, the sports scoreboard. Collect the findings and build the definition from what they brought back. Close by connecting to Chapter 1: every AI example they discussed last week ran on data collected from somebody.

Period 2 is the conceptual heart. Run the pixel-grid activity in the first twenty-five minutes so students discover the image-as-numbers idea themselves rather than being told. Spend the remaining time on text codes and the sound waveform. Do not go anywhere near binary or bits — that is not in the Class 6 scope, and it will lose the class.

Period 3 is a working period. Pick a question that genuinely varies across your class and that nobody will be embarrassed to answer. Travel mode works everywhere. Favourite mid-day meal item, number of siblings, and month of birth also work. Avoid anything touching family income, caste or phone ownership.

Period 4 is drawing, and it is slower than you expect. Thirty-six students choosing a scale on squared paper takes twenty minutes on its own. If the maths chapter has already been taught, this halves. If it has not, you will lose the analysis period to catching up — which is the exact reason for sequencing the maths chapter first.

Period 5 is the one teachers skip when they run short, and it is the one that must survive. Put four charts on the board — one with no title, one with a truncated axis, one pie with nine slices, one with uneven scale gaps — and let groups diagnose them. Then turn the question on their own data. Our CT and AI teacher training programme treats this period as the assessable core of the chapter, not the optional extra.

Unplugged activities

Activity 1 — The pixel grid: a picture hidden in numbers

Time: 25 minutes. Materials: a printed or hand-drawn 8×8 squared grid per student, pencils, and one number key on the board. No devices.

1. Draw an 8×8 grid on the board and number the rows 1–8 and the columns 1–8. Ask students to copy it, or hand out the printed sheet.
2. Write the shading key on the board: 0 = fill black, 128 = shade grey, 255 = leave white. Nothing else is explained.
3. Read out the 64 numbers row by row, slowly, twice. Choose a simple shape — a smiling face, a diya, a cricket bat, the first letter of the school’s name.
4. Students shade each cell as the number is called. Tell them nothing about what they are drawing.
5. When the last row is done, ask them to hold up their grids. The image appears across the room simultaneously. Let the reaction happen.
6. Now ask: how many numbers did it take? Sixty-four. A real photograph is a grid of millions.
7. Extension for a strong class: read the same 64 numbers in the wrong order — column by column instead of row by row — and let them see the picture break. The numbers alone are not enough; the arrangement is part of the data.

Discussion question: If a picture is only numbers, could a machine change those numbers to change the picture? What would that mean for a photograph used as evidence?

Activity 2 — The class survey, and the wrong conclusion

Time: 35 minutes, spanning periods 3 and 5. Materials: notebook, squared paper, blackboard. No devices.

1. As a class, agree on one question with four to six clear options. Write the exact wording on the board. Argue about the wording for two minutes — this is not a delay, it is the lesson.
2. Agree the rules before collecting: what happens if someone uses two modes, what counts as “Other”, what to do about absentees.
3. Two students run the tally at the board while everyone answers by raising hands, one option at a time. Everyone else tallies in their notebook.
4. Convert to a frequency table. Add the total row. If the total does not match the number of students present, stop and find the error before continuing.
5. Record three pieces of context under the table: today’s date, the number of students present, and the number absent.
6. Each student draws a bar chart from the table. Apply the five-point checklist to a neighbour’s chart, not their own.
7. In period 5, each group writes two sentences on a slip: one thing the data definitely shows, and one thing someone might wrongly conclude from it. Read the wrong conclusions aloud and let the class judge each one.

Discussion question: What could someone wrongly conclude from this data? Push until they produce at least three distinct errors: over-generalising to other classes or schools, confusing correlation with cause, and forgetting the absentees.

Activity 3 — Chart hospital

Time: 20 minutes. Materials: four deliberately broken charts drawn on chart paper or on the board; one diagnosis slip per group.

1. Prepare four patients in advance. Patient A: a bar chart with no title and no axis labels. Patient B: monsoon rainfall in mm with a y-axis starting at 60. Patient C: a pie chart of nine favourite subjects. Patient D: a bar chart whose y-axis reads 0, 5, 10, 50, 100 at equal spacing.
2. Groups of four rotate through the charts, three minutes each, writing one diagnosis per chart.
3. Each group presents the single worst problem they found and the specific fix.
4. Rank the four patients from most misleading to least, as a class, with a reason for the ranking.
5. Finish by asking which of the four errors could have been an honest mistake and which looks deliberate.

Discussion question: A chart cannot contain a false number and still be a chart of the real data — so how can a chart with entirely true numbers still mislead someone?

Worksheet

Use the following table for questions 5 to 8.

Part A — Recall

1. Name the four types of data listed in your syllabus, and give one example of each from your school.
2. What does a computer store instead of the letter M?
3. What is a pixel?
4. Write two things that every bar chart must have, apart from the bars themselves.

Part B — Apply (use the rainfall table)

5. How much rain fell on Thursday?
6. Which day had the highest temperature, and what was it?
7. What was the total rainfall for the week?
8. On the two wettest days, what happened to the temperature? Write one sentence.
9. A student records the class’s favourite mid-day meal item as: rice 14, roti 9, khichdi 7, dal 6. Draw a labelled bar chart with a scale from 0 to 16 in steps of 2. Write the value above each bar.
10. Convert the same mid-day meal data into a frequency table with a total row. How many students were asked?

Part C — Think

11. A newspaper prints a bar chart of the runs scored by two batters. The bars show 48 and 52, but the taller bar looks three times the height of the shorter one. What has the newspaper done, and why is it misleading?
12. A student surveys the six friends sitting near her and writes: “Class 6 prefers khichdi.” Give two reasons why this conclusion is unsafe.
13. Your school wants to know which subject students find most difficult. You may ask only 20 of the 200 students in the school. How will you choose the 20? Give one reason for your method.
14. A voice note from your friend is 30 seconds long. Explain, in two sentences, how it is stored as data.
15. Someone shows you a pie chart with eleven slices. Suggest a better chart and explain why.
16. A school claims: “Attendance improved after we started the new morning assembly.” The data covers one week before and one week after. What could someone wrongly conclude, and what extra data would you want?

Answer key

1. Numbers, text, images, sound. Examples: attendance count; a student’s name on the register; the school photograph; the morning bell recording.
2. A number code that has been agreed to mean M. (Accept: a number from a standard code table.)
3. One of the tiny squares that make up a picture, storing numbers for its brightness or colour.
4. Any two of: title, labelled axes with units, an even scale starting at zero, equal bar widths, value labels.
5. 45 mm.
6. Wednesday, 34 °C.
7. 12 + 68 + 0 + 45 + 5 + 90 = 220 mm.
8. Saturday (90 mm) and Tuesday (68 mm) were the wettest, and the temperature was lowest on those days — 26 °C and 27 °C. Accept any sentence noting that temperature fell as rainfall rose. Do not accept “rain causes low temperature” as a proven claim.
9. Chart with title (for example “Favourite mid-day meal item, Class 6”), y-axis “Number of students” 0–16 in steps of 2, x-axis “Meal item”, four equal bars of heights 14, 9, 7, 6 with values written above.
10. Rice 14, roti 9, khichdi 7, dal 6; total 36 students asked.
11. It has cut the y-axis so the scale does not start at zero (a truncated axis). The real difference is 4 runs, roughly 8 per cent, but the picture suggests one batter scored three times as much.
12. Any two of: only six people were asked out of the whole class; they were all her friends and friends tend to be similar; the rest of the class and any absentees were never asked; a different day’s meal might give a different answer.
13. Any reasonable method that spreads the sample across the school — for example, choose four students from each of the five classes, or pick every tenth name from the school register. Reason: asking only one class or only one’s own friends would describe that group, not the school.
14. The microphone measures the sound thousands of times each second and each measurement is written down as a number. The 30-second note is therefore stored as a very long list of numbers, which are played back in order to recreate the voice.
15. A bar chart. With eleven slices the angles are too similar to compare and the labels will not fit; bars can be ranked by height and read against a scale.
16. Wrong conclusion: that the assembly caused the improvement. Something else may have changed — the weather, an exam week, a festival, illness. Extra data wanted: several months before and after, attendance in a class that did not change its assembly, and the reasons recorded for absence.

Common misconceptions

Assessment

At this stage the CT and AI curriculum is competency-based and largely internal. Assess what students do — the survey they ran, the table they built, the chart they drew, the wrong conclusion they identified — rather than definitions they can recite. Do not invent a marks weighting; the handbook does not specify one for this chapter, and schools should follow whatever internal scheme their assessment policy sets.

Four things are worth assessing:

• Can the student classify a piece of data into the correct type and justify it?
• Is their frequency table complete, totalled and internally consistent?
• Does their chart pass the five-point checklist without prompting?
• Can they state one thing the data does not show?

Interdisciplinary links

Class 6 carries 40 hours of interdisciplinary projects, and Chapter 2 is the single most reusable chapter in the AI block — almost every project needs data collected, organised and presented. Use it as the service chapter it is.

• Mathematics — Data Handling and Presentation. The direct partner. Teach the NCERT chapter first, then let the AI chapter supply the real datasets the maths chapter needs. Project: the class runs a survey the maths teacher marks for accuracy and the AI teacher marks for honesty.
• Science. A weekly weather log — temperature in °C and rainfall in mm, recorded at the same time each day for a month. Students discover for themselves why the reading time must be fixed, which is a data-quality lesson no lecture delivers.
• Social science. The Census of India as the country’s largest data collection exercise. What is counted, what is not, and who might be missed. This connects the sampling idea to something at national scale.
• Languages. Word-frequency counts in a poem or a news report, presented as a bar chart. Then the harder question: does the most frequent word tell you what the passage is about?
• Art and design. Pixel art on squared paper, exchanged as number grids between pairs. One student encodes, the other decodes. Extend to a 16×16 grid with three grey levels.
• Physical education and health. Heart rate before and after skipping, across the class. A genuine measurement dataset with natural variation, and an obvious wrong conclusion waiting to be made.

Each of these leans on the computational thinking half of the year too. Deciding what to record and what to ignore is abstraction; spotting that rainfall rises and temperature falls is pattern recognition, which Chapter 3 takes up directly; breaking a survey into agree-the-question, collect, check, present is decomposition. For the full picture of how the strands interlock, see the CT and AI pillar guide.

Frequently asked questions

Do students need a computer or a spreadsheet for this chapter?

No. The Class 6 AI content requires no coding, no Python and no AI tool. All five hours can be taught with a blackboard, squared paper and the students in the room. If your school has computers, a spreadsheet can speed up chart drawing in period 4, but it is not required and it should not replace hand-drawing at least one chart.

Should we teach the maths chapter Data Handling and Presentation before this one?

Yes, and this is the sequencing decision most schools get wrong. The AI chapter assumes tally marks, frequency tables and bar graphs are already familiar. If they are not, you will spend three of your five hours on drawing mechanics and lose the analysis. Coordinate with the maths teacher at the start of the term. CBSE does not mandate this order; it is a practical recommendation.

Is binary or ASCII part of the Class 6 syllabus?

No. The syllabus line says introduction to data types, not number systems. It is enough that students understand a letter is stored as an agreed number code and a picture as a grid of numbers. Introducing binary at Class 6 adds difficulty without adding understanding.

How is this chapter assessed?

At this stage the CT and AI curriculum is competency-based and largely internal to the school. Assess the artefacts students produce: the frequency table, the labelled chart, and their answer to what the data does not show. The handbook does not prescribe a marks weighting for individual chapters, so follow your school’s internal assessment policy.

What if the class survey produces embarrassing or sensitive answers?

Choose the question carefully in advance. Travel mode, mid-day meal preference, birth month and number of siblings are safe in almost every classroom. Avoid family income, caste, religion, phone or vehicle ownership, and anything about marks. The point of the activity is the method, not the topic, so pick the least sensitive topic that still varies across the class.

How does this chapter connect to Class 7 and Class 8?

Directly. Class 7 introduces machine learning, where a model learns from training data; students who already ask who was asked and what was left out will understand biased training data immediately. Class 8 covers the AI project lifecycle, in which data collection and preparation is an explicit stage. Chapter 2 is where that vocabulary starts.

PiyushAI is founded by Piyush Wairale (IIT Madras) and works with schools on Redefining Education and Exam Preparation with AI. We build classroom-ready teaching guides for the CBSE Computational Thinking and AI curriculum, run CT and AI teacher training, and maintain a plain-language AI glossary and an AI literacy track for students. If you are planning the 2026–27 rollout, start with the school AI readiness score.

Related: Class 6 AI Chapter 1: Introduction to AI · Chapter 3: Pattern Recognition and Decision Making · Chapter 4: AI Ethics and Digital Responsibility · Class 6 CT and AI syllabus · Class 5 syllabus · Class 7 syllabus · Pattern recognition module · Algorithmic thinking module

Source: CBSE Computational Thinking & Artificial Intelligence handbook, Class 6 (2026–27), cbseacademic.nic.in.