What are the best coding classes in Dundee?
Scotland's Census 2022 put about 148,100 usual residents in Dundee City, with 67.1% of them aged 15 to 64 against 64.6% across Scotland. The research ship Discovery, built in the city, carried the British National Antarctic Expedition of 1901 to 1904 and came home to Dundee in 1986. Our Dundee learners, aged anywhere from six to sixty-seven, study with us over live video: one to one, or in a group of five to ten set by level, with teachers in India working to UK times. A first lesson is free, and after that a group place is USD 100 a month while one-to-one teaching is USD 150.
The Met Office station at Leuchars, just across the Firth of Tay, has recorded every month since January 1957. Suppose a program has to forecast next month's average daytime high. The lazy guess, that next month will be like this one, misses by 2.296 degrees on average over 2000 to 2025. The long-run average for that calendar month misses by 1.113. Add a fifth of last month's departure from normal and the error falls to 1.009. Tune that fraction by peeking at the years being tested and it seems to fall to 0.969, but that number is a cheat, and so is any average that quietly includes the future. Our teenagers learn to build a forecast and, more importantly, to test it fairly.
Facts last verified 22 September 2026. Teaching is online; no Dundee branch is claimed.
Let the learner's interests choose. Every course begins with a live lesson at no charge, booked without any card.

Python for younger learners, including little prediction games where the computer guesses tomorrow and keeps score of how wrong it was.
See the syllabus
Reading a real weather file, building forecasts and scoring them honestly, the route to the Leuchars project on this page.
See the syllabus
Machine learning for teenagers, where every model is compared first with a simple baseline and tested on data it never saw.
See the syllabus
Time series, forecasting and error measures for adults who plan with data at work.
See the syllabusBrowse the course atlas for more than one hundred options and use the coding roadmap to check prerequisites.
The four we are known for
These run underneath everything above. Every one is live and online, placed by ability rather than by age, and the first class is free.

Python, web and AI projects where the learner still owns the thinking.
See the syllabus
Automate the work you already do, then let AI carry part of it.
See the syllabus
Train a model, read what it learned, and be able to say why it is wrong.
See the syllabus
Run AI coding agents on real work without losing control of the codebase.
See the syllabusFrom National Records of Scotland's first results of the 2022 census. Counts are rounded to the nearest hundred and printed as released.
| Measure | Dundee | Scotland |
|---|---|---|
| Usual residents, rounded | 148,100 | 5,436,600 |
| Usual residents in 2011 | 147,268 | Not compared |
| Households, rounded | 69,900 | Not compared |
| Households in 2011 | 69,193 | Not compared |
| Aged 20 to 24, rounded | 13,900 | 337,000 |
| Share aged 15 to 64 | 67.1% | 64.6% |
| Share aged 65 and over | 18.1% | 20.1% |
| Residents per square kilometre | 2,476.6 | 69.8 |
About 148,100 people lived in Dundee City in 2022, against 147,268 in 2011, in about 69,900 households against 69,193.
Two thirds of residents, 67.1%, were aged 15 to 64, above the Scottish share, while 18.1% were 65 or over against 20.1%.
At 2,476.6 residents per square kilometre, the council area is tightly drawn around the city, with Broughty Ferry inside it and Monifieth just beyond.
National Records of Scotland also reports that 21.8% of Dundee households had dependent children in 2022. Our learners here cover every stage: a P3 pupil in Broughty Ferry making a first game, an S5 student in Menzieshill working towards Higher Computing Science, or an adult in Lochee learning Python for a new job, each in a group at the right level.
From Dundee Heritage Trust, which has cared for the ship since she came home.
The Trust describes Discovery as the first British commissioned scientific research ship, built in Dundee for the British National Antarctic Expedition under Captain Scott.
The crew compiled the South Polar Times on board, one of the publications the Trust now celebrates in its polar collection.
After voyages all over the globe, including oceanographic expeditions in the 1920s, Discovery returned home to Dundee in 1986.
We have no connection with Dundee Heritage Trust. Discovery belongs on this page because she was built to do science at sea, where careful records and honest measurement were the whole point of the voyage. The project below keeps that spirit on dry land: a weather record kept for seventy years, and a test of whether anyone can predict it better than the obvious guess.
Where the records come from
The Met Office publishes monthly records for a set of long-running stations as free text files. Leuchars, in Fife, about 11.6 kilometres from Dundee city centre, has monthly maximum and minimum temperatures, rain and sunshine from January 1957 to the present, with the most recent months marked provisional.
Every forecast here is made in order, month by month, using only the months before it. That rule, called walk-forward testing, is what makes the scores fair.
| Method | Average miss | Fair? |
|---|---|---|
| This month again (persistence) | 2.296 C | Yes |
| Same month last year | 1.375 C | Yes |
| Long-run average for the month (climatology) | 1.113 C | Yes, using earlier years only |
| Climatology plus 0.2 of last month's anomaly | 1.009 C | Yes, 0.2 chosen on 1970 to 1999 |
| Climatology plus 0.4 of last month's anomaly | 0.969 C | No, 0.4 picked by looking at the test years |
| Climatology from all years to 2025 | 1.032 C | No, it averages in the future |
Before any clever model, score the obvious guesses. A method that cannot beat the long-run monthly average is not worth using, however sophisticated it looks.
Forecast January 2000 from data up to December 1999, then February from data up to January, and so on. The model never sees the month it is predicting.
The weight on last month's anomaly was chosen on 1970 to 1999, before the test began. Choosing it on the test years flatters the result.
Month by month, the long-run average was the closest of four methods only 54 times out of 312, fewer than simply repeating last month, which won 57. Yet its average miss is less than half the size. Persistence is sometimes spot on and sometimes badly wrong, especially at the turn of the seasons; climatology is rarely exact and never far away. Average error, not a count of wins, is the fair score.
Sunshine is harder to beat. For monthly sunshine hours, the long-run average misses by 20.426 hours, and adding a fifth of last month's anomaly improves that only to 20.258, while persistence misses by 34.773. A learner who reports that honestly, that the extra method barely helps for sunshine, has learned more than one who only reports the temperature success.
Learned on seventy years of Fife weather, then used on sales forecasts, energy demand, exam predictions, sports models and every machine learning project.
| Rule | At Leuchars | What breaks without it |
|---|---|---|
| Beat a baseline | Every method scored against climatology first | Celebrating a model that the monthly average beats |
| Only the past | Each month forecast from earlier months alone | A score no real forecaster could ever achieve |
| Tune before testing | Weight chosen on 1970 to 1999, tested on 2000 to 2025 | A parameter fitted to the answers |
| Average the errors | Mean absolute error over 312 months | Being fooled by a method that wins often but misses badly |
| Report the losses too | Sunshine barely improves, and the page says so | A success story built by leaving results out |
The leaky climatology in the table shows how easily the second rule breaks. Averaging each calendar month over all the years up to 2025, including the year being forecast, cuts the error from 1.113 to 1.032 without any extra skill. The code looks innocent. Only asking what the program could have known at the time exposes it.
Guessing tomorrow's temperature every day for a fortnight, writing the guesses down first, and adding up how far off they were.
The Leuchars file read in Python, four forecasting methods, walk-forward testing and a parameter frozen before the test period.
Forecast evaluation at work, with baselines, holdout periods and error measures agreed before anyone looks at results.
We are not connected with the Met Office, National Records of Scotland, Dundee Heritage Trust or Dundee City Council. The station record and census figures are published openly; the forecasts, errors and comparisons on this page are our own work.
These bands are a loose guide; the free lesson shows where a learner actually belongs.
Making a prediction, writing it down before the answer arrives, and measuring how close it was.
Coding for KidsMental Maths for KidsUsing past months to guess the next one in code, and comparing two guessing rules fairly.
Python and AI for KidsMaths Through CodingBaselines, anomalies, walk-forward loops and error measures on seventy years of real weather records.
Data Science for TeensAI and Machine Learning for TeensTime series methods for adults, with evaluation designed before modelling so the results can be trusted.
Data Analysis CourseStatistics & ProbabilityBecause a forecast nobody has tested fairly is only a guess with decimals.
Ask an assistant to predict next month's weather or next quarter's sales and it will produce confident numbers. A learner who has scored forecasts walk-forward knows to ask what baseline it beats, which data it was tuned on, and whether any future information leaked into the test. Those three questions catch most of the misleading results in data science.
Every serious AI project is evaluated the same way: a baseline, a held-out test set that the model never sees during training, and choices frozen before the test is run. Learning those habits on a weather record from across the Tay is easier than learning them on a large model, and the habits are identical.
So a Dundee teenager should still learn to program in 2026, in a city that built a ship to do science at the bottom of the world: predictions are cheap, and knowing which ones deserve belief is the valuable skill. The longer argument is in why coding is worth learning in 2026.
A dark winter evening in Dundee is no time to cross the city for a lesson. Online, the lesson comes to the learner.
A learner in Whitfield and another in Menzieshill can join the same class without either travelling.
P1 to P7, S1 to S6, then National 5, Higher and Advanced Higher: lessons follow the names Dundee schools use, and every lesson is in English.
The free session teaches a real task and closes with a recommended level, course and weekly time. No payment card is requested.
Classmates at the same stage from Dundee, elsewhere in the UK and overseas, so each level has a workable time slot.
A fixed pair of lessons each week, roughly eight a month, with holidays and exam leave planned with the teacher well in advance.
Indian time is constant all year, which puts it four and a half hours ahead of Dundee in the summer and five and a half in winter, so after-school and evening lessons in the UK fall inside our teachers' working day.
Across Tayside and Fife
Families in Monifieth, Carnoustie, Newport-on-Tay or Forfar join the very same groups, because lessons are online and groups are formed by level.
Start with a free lesson; after that, one monthly fee.
A complete lesson with nothing to pay, closing with a suggested level, course and weekly slot.
Close to eight live lessons a month, shared with five to ten learners at the same stage.
Close to eight live lessons a month, taught to your learner on their own.
Outside India the fee is the same for everyone and set in US dollars, so Douglas and the West End pay alike, and no prices appear in pounds. Billing starts only once the free lesson has fixed a course and a time; pausing, missed lessons and switching between group and private teaching are explained on the pricing page.
Rated 4.9 across 547 Google reviews. These are real reviews, reproduced as written.
★★★★★
"The one step solution for my son. Modern Age Coders make learning coding so simple that kids love it. The teachers explain complex concepts clearly with practical exercises and interactive content."
Ria Mukherjee
Parent
★★★★★
"Modern Age Coders has been a game-changer for me. I struggled to grasp IT concepts and coding before joining, but their classes transformed everything. I can now confidently write complex programs with ease."
Samriddha Mondal
Student
★★★★★
"One of the most wonderful education centres out there. Education is not limited to school syllabus but focuses on skill development."
Vansh Agarwal
Student
★★★★★
"My child Dhairya is really enjoying the Modern Age Coders classes. This is his first online class and he eagerly looks forward to it. I can already see his improvement, and the teachers are very cooperative."
Sonam Oswal
Parent of Dhairya
★★★★★
"Modern Age Coders have wonderful teachers who teach in a clear, easy and practical way. The teacher boosts students' confidence and inspires them to learn without hesitation."
Sonu Goyal
Parent
★★★★★
"I highly recommend this computer coding class! The teachers are incredibly knowledgeable and passionate about coding."
Ritu Kedia
Parent
The first task fits the learner: a guess-and-check prediction game for a young child, a short Python program that averages past months for a beginner, or the Leuchars record and a walk-forward test for a teenager ready for real data.
The questions we hear most from Dundee families.
Scotland's Census 2022 counted about 148,100 usual residents in Dundee City, rounded to the nearest hundred, compared with 147,268 in 2011, living in about 69,900 households.
In 2022, 67.1% of Dundee residents were aged 15 to 64 against 64.6% nationally, 18.1% were 65 or over against 20.1%, and there were 2,476.6 residents per square kilometre against 69.8.
Testing a forecasting method in time order, so that each prediction is made using only data from before the thing being predicted. It stops a model from being scored with information it could never have had.
For monthly average daytime highs from 2000 to 2025, the long-run monthly average plus a fifth of last month's departure from normal missed by 1.009 degrees on average, ahead of the plain average at 1.113 and last month's value at 2.296.
The Met Office publishes monthly records for Leuchars, in Fife, as a free text file, running from January 1957. The latest months are marked provisional, and we leave them out of the tests.
A ship built in Dundee that Dundee Heritage Trust describes as the first British commissioned scientific research ship. She carried the 1901 to 1904 British National Antarctic Expedition and returned to Dundee in 1986. We are not connected with the Trust.
After school or in the evening on weekdays, or at weekends, at a UK time we settle in the free lesson. India is four and a half hours ahead of Dundee in summer and five and a half in winter.
No. We have no Dundee centre and no premises anywhere in the UK; every lesson is live online. A learner needs a computer with sound and a stable connection, and our phone number is Indian.
The first lesson is free. After it, a group place is USD 100 a month for two live lessons a week, about eight a month, with five to ten learners, and one-to-one lessons on the same pattern are USD 150 a month. Course, format and time are agreed before any charge.
By level, pace and aims rather than age or postcode, with five to ten learners at one stage. If no group suits the learner's week, we offer one-to-one lessons.
Up the coast, the Aberdeen page fits every place name in the city into a few kilobytes, and to the south Edinburgh works out why its summer nights never quite go dark. The Scotland guide covers the Curriculum for Excellence, with exam support on the National 5 and Higher Computing Science pages.