AI and Machine Learning Workshop at Biyani Girls College: A Detailed Learning Experience for BCA Students

Ask ten people what machine learning is and you’ll get ten grand answers. Mine would be smaller. You collect data, tidy it, look hard at it, train a model on it, and then try to catch the model being wrong. For students of computer applications, learning that routine early gives later study in programming and data science something solid to stand on.

That was the thinking behind the 100-hour AI and Machine Learning workshop run by the Department of Information Technology at Biyani Girls College from 1 September to 15 October 2026. It was held offline, in association with IBM and Edunet Foundation, for BCA second-year students in their third semester.

Mr. Ronak Patidar was the trainer. Python, numerical computing, working with data, charts, machine learning ideas and responsible AI all made the list. At the end, students presented a capstone project, and explaining your work to people who haven’t seen it is its own skill.

Theory and tools ran side by side. Students came away having seen the whole chain once, raw data at one end and a tested model at the other.

AI and Machine Learning Workshop: Programme Overview

The workshop sat alongside regular classes and didn’t try to replace them. It also didn’t stick to one topic. The sessions moved from handling data right through to the basic stages of machine learning.

Quick facts: the host was the Department of Information Technology at Biyani Girls College. Teaching time came to 100 hours, all of it offline, spread across 1 September to 15 October 2026. Mr. Ronak Patidar trained the group, IBM and Edunet Foundation were the associated organisations, and the participants were BCA second-year students in their third semester.

Programming libraries got plenty of attention, mostly for what they do in data analysis and model building. Another habit got airtime too, and it isn’t flashy: double-checking results, and thinking about who an AI system could end up affecting.

1. Python Programming: The Foundation for AI and Machine Learning

Python shows up in automation scripts, in web back ends, in data analysis. It reads almost like English with brackets, and its libraries are so numerous that a beginner reaches a working result sooner than expected. I think that’s why it’s such a common first language for data work.

Most of the legwork in an AI workflow ends up being Python code. Loading files. Fixing the mess that came with them. Running the sums, drawing the charts, calling the library that builds the model.

Say a student wants to analyse a dataset. First comes the unexciting part: import the data, see how it’s laid out, find the missing values. Python lets those steps be written down once and repeated, so the process doesn’t depend on someone remembering what they clicked last Tuesday.

That’s why Python mattered here beyond the language itself. It joined programming basics to real analysis, and it readied students for the tools that followed.

2. NumPy: Understanding Numerical Computing

NumPy is where the arithmetic lives. A lot of data science is the same sum done a million times, and NumPy does it quickly using arrays. The workshop included it among its data tools.

Indexing, reshaping arrays, element-wise arithmetic and aggregation are the everyday features, along with general mathematical calculation.

Picture a table of marks for different subjects. An average, a subject-by-subject comparison, a column total: each is a short NumPy call, with no hand-written loops.

After that it’s easier to see how numbers get stored and processed before any modelling starts.

3. pandas: Preparing and Analysing Data

A model learns from whatever it’s given. If that data has gaps, mismatched formats or the wrong data types, the analysis can go wrong without any warning at all.

pandas came next, the Python library for tabular data. Two structures do most of its work: Series and DataFrame.

In practice that means importing a dataset and looking it over, filtering rows or columns, hunting down missing values and deciding what to do with them, joining tables and grouping records for descriptive statistics. Before any of that, you check the data quality.

Imagine a sheet of student attendance. With pandas you can filter the records, total up the attendance figures and find the entries nobody filled in.

I’d argue this is the unglamorous half of data science, and probably the more important half. Which algorithm you pick matters. The quality and suitability of your data matters at least as much.

4. Data Visualisation Using Matplotlib and Seaborn

Tables are good for detail and bad for patterns. A chart shows the pattern at a glance. That’s why visualisation is a standard part of exploratory data analysis.

The workshop used Matplotlib’s pyplot interface and Seaborn to present and explore data.

Line plots, bar charts, scatter plots and histograms all fall within Matplotlib’s range. Those cover comparing values, checking distributions and showing how things change across observations.

Seaborn sits one level higher. It offers statistical visualisation tools and works directly with pandas DataFrames, which suits exploring how variables relate to each other.

Bar charts suit comparing categories. Want to see how a measurement is spread across a group? Histogram. Suspect two variables move together? Scatter plot.

A badly drawn chart can mislead you without anyone intending it. Clear labels, sensible scales and appropriate visual choices stop a chart from misleading anyone. And a relationship on a chart doesn’t prove that one variable causes the other to change.

5. Artificial Intelligence, Machine Learning and Deep Learning

AI, ML and deep learning get used interchangeably, which is understandable and wrong. Related, yes. Same, no.

Artificial Intelligence

Think of AI as the umbrella term for systems built to do things we associate with language, spotting patterns, planning or helping people decide. How they do it varies from one task to the next.

Machine Learning

Machine learning is one route under that umbrella. Nobody writes out every rule. The model looks at examples and picks up patterns it can then use to predict or classify.

How good it turns out depends on three things, in order: the data, the method, and how honestly someone checked the results.

Deep Learning

Deep learning goes a step further, using neural networks with many layers. It’s behind image recognition, speech processing and many language tasks.

Stack them and you get nesting: deep learning inside machine learning, machine learning inside AI. That picture helps students place each term.

6. Types of AI and Machine Learning

Next came classification, both of AI and of machine learning. The first tells you what a system can do. The second tells you how a model learns.

AI Classifications

The workshop covered narrow AI, and then two ideas that are still hypothetical: general AI and superintelligent AI.

Narrow AI is built for specific tasks within a defined scope. General AI would handle a broad range of intellectual tasks, though it remains hypothetical. Superintelligent AI is the hypothetical idea of something that surpasses human intellectual ability across a wide range of areas.

It’s worth being clear that general and superintelligent AI are conceptual categories. They don’t describe established, generally available systems.

Machine Learning Approaches

Four common approaches were covered.

Supervised learning is the one with an answer key: the model learns from labelled examples, then predicts outcomes or classifies new data. Take the labels away and you’re in unsupervised territory, where the job is finding patterns or structure on its own. Semi-supervised learning sits in between, with a mix of labelled and unlabelled data. Reinforcement learning is different again. An agent acts in an environment and learns from the feedback its actions earn.

No approach wins everywhere. Pick by the problem, by the data you actually have, and by what you need at the end.

7. Introduction to scikit-learn and Model Development

After the data preparation comes scikit-learn, the Python library that supplies standard machine learning tools so nobody builds them from scratch. Classification, regression, clustering, preprocessing, model selection and evaluation all live in it.

The workshop used it to show a general workflow, and the order matters. Start with the problem: what are you trying to answer, and what will anyone do with the answer? Then look at the data, its sources, formats, missing values and likely weak spots, and prepare it by cleaning errors and applying suitable transformations. Next, split the training data from the test data. Keeping the test set separate is what lets you estimate how the model copes with examples it has never seen.

Now train the model: choose a suitable method and fit it to the training data. Evaluate it with metrics that suit the task, and look hard at where it gets things wrong. Last, tell people what you found, limits and risks included.

That’s seven steps, and the algorithm is only one of them. Data quality, evaluation and clear communication weigh as much.

8. Capstone Project Presentation: Teamwork and Technical Communication

The workshop ended with a capstone project presentation, done in teams of two to four students.

Students got to pitch a project idea and explain the technical side to an audience. A good presentation walks through the problem and who it’s for, then the approach, the tools and the results. After that comes the honest part: what’s still weak, and what would make it better.

Doing this in a team trains collaboration, planning and plain-spoken technical explanation. Those pair well with programming and analysis, particularly in academic projects or group technology work.

One caution. The workshop report records the capstone presentation as an activity but doesn’t give individual project titles or results. So it should be read as an opportunity to communicate project work, and it isn’t evidence of any particular project achievement.

9. Responsible and Ethical Use of Artificial Intelligence

AI output can be useful. It isn’t automatically accurate, fair or appropriate. Responsible practice starts with asking what a tool can’t do, and who might be affected when it gets things wrong.

Five ethical issues came up.

Fairness and Bias

Training data carries biases, and models inherit them. Does the dataset actually represent the people or situations the system will meet? And where it makes sense, have the errors been examined?

Privacy and Data Protection

Treat personal and confidential data with care. Use it only for suitable purposes, and keep sensitive material out of AI tools unless you have permission.

Transparency and Accountability

Anyone relying on an AI system deserves to know its purpose, what its outputs will be used for and where it falls short. And if it feeds into something important, somebody specific still owns the decision.

Reliability and Human Oversight

Wrong answers come out of AI tools in explanations, predictions and code alike. Check outputs against reliable information. If a wrong result could do real damage, a human has to stay in the loop.

Academic Integrity

Submit only what you understand. Verify any AI-generated explanation or code, and follow your institution’s rules on acknowledging sources and AI help.

Strip it down and all five say the same thing: question the output before you trust it.

10. Key Learning Outcomes of the Workshop

The programme introduced foundational ideas. Further study and practice can take students well past them.

The documented outcomes cover a lot of ground. Students should be able to explain how AI, machine learning and deep learning relate, and recognise the common machine learning approaches while knowing that different tasks need different methods. They should see how Python libraries support numerical analysis, data preparation and visualisation, and be able to describe the main stages of a machine learning workflow. They should understand why test data, suitable evaluation metrics and honest reporting matter. On the ethics side: fairness, privacy, transparency and responsible AI use. And they should be able to put technical ideas across in a team presentation.

Treat them as a starting line. Deepening them means writing programs, analysing suitable datasets, and evaluating models on small projects with clear limits.

11. Recommendations for Continued Learning

The report doesn’t stop at the workshop. It recommends follow-up work.

Mostly it comes down to practice. Regular exercises keep Python fundamentals and individual libraries fresh, and small projects on suitable datasets give students somewhere to try data preparation, visualisation and introductory modelling.

There’s a note about record-keeping too: students should write down their data preparation steps, coding decisions, model results and limitations. It’s a good habit. Careful analysis follows from it, and so does work that someone else can actually review.

Evaluation metrics need more practice too, since that’s how students learn to read model performance and recognise common modelling errors. Privacy, academic integrity and verifying AI output need steady attention alongside.

Do that for long enough and introductory knowledge becomes a practical foundation.

Conclusion: Strengthening AI and Data Science Foundations at Biyani Girls College

The Department of Information Technology at Biyani Girls College ran its AI and Machine Learning workshop to give BCA students a first working acquaintance with the concepts and tools behind AI and data science.

The stated length was 100 hours, between 1 September and 15 October 2026. IBM and Edunet Foundation were the partners, and Mr. Ronak Patidar was the trainer.

On the technical side the topics were Python, NumPy, pandas, Matplotlib.pyplot, Seaborn and scikit-learn. On the conceptual side they were AI and machine learning classifications, deep learning and responsible AI. The capstone presentation added teamwork and practice in explaining technical ideas.

A workshop is a start. Practice is what carries it forward. The students who keep working with code, datasets and small projects are the ones likely to get the most out of it, and to be ready for more advanced study.

Through academic activities that bring in emerging technologies, Biyani Girls College gives students room to explore important areas of information technology alongside their regular coursework.

Frequently Asked Questions (FAQs)