WBCHSE HS Artificial Intelligence & Data Science

Expect a written theory paper mixing multiple-choice, short-answer and longer descriptive questions, a laboratory practical in Python, and a project assessed through viva-voce. The practical typically asks you to load a dataset, clean it, compute summaries and produce a labelled chart. Confirm the current marks distribution and time allowed on the official WBCHSE site, which publishes the scheme each session.

This page does not quote question counts, marks or duration, because they change between sessions and any figure here would quickly become wrong. The West Bengal Council of Higher Secondary Education publishes the current scheme of examination, full marks and time allowance for each subject on its official website and in its annual notification. Always rely on the latest circular for your session.

Compartmental and repeat examination rules are set by the Council, not by schools or coaching centres, and they are revised from time to time. Whether you may reappear in a single paper, and in which year, depends on the regulations in force. Check the WBCHSE notification for your session, or ask your school's subject coordinator, who receives the circulars directly.

Moderate. The theory paper is largely conceptual - definitions, comparisons between methods, tracing short Python snippets, interpreting plots and metrics - so precision of vocabulary matters more than long derivations. The practical and project are where marks are most easily lost: unlabelled graphs, unhandled missing values, sloppy file paths, or a project you cannot explain during the viva.

Start with Python fundamentals, because every later topic assumes you can write a loop and call a library function. Move to data handling and visualisation with pandas and Matplotlib, then descriptive statistics and probability, then machine learning. NLP and computer vision sit comfortably after that. Weave ethics through the whole year rather than leaving it as a final chapter to memorise.

Less than students fear. You need mean, median, mode, standard deviation, correlation and basic probability - all computable in code. For neural networks, an intuitive grasp of how a gradient shows the direction of error reduction is enough; formal calculus is not examined. Comfort with percentages, averages and reading a table of numbers matters far more than algebra.

No. Everything in the practical syllabus runs on free tools: Python with NumPy, pandas and Matplotlib, installed locally through Anaconda or used in a browser-based notebook environment. A modest laptop is sufficient; datasets used at this level are small. Check with your school about the laboratory setup, since practical examinations are conducted there.

Rote-learning definitions without being able to apply them, confusing supervised with unsupervised learning, muddling classification with regression, and quoting an accuracy figure without saying what it measures. In practicals, the usual culprits are missing axis labels, ignoring missing values, and copying code that will not run on a fresh dataset. In projects, it is a model you cannot justify.

Be ready to explain four things without notes: where the data came from and what one row represents, what cleaning decisions you made and why, why you chose that particular model, and what your result means in plain language. Also know your project's limitations. Examiners rarely ask about advanced theory; they probe whether you genuinely built and understood the work yourself.