BITS Pilani Data Analytics Internship Project Scoping

November 13, 2026

Web Stories

A data analytics project fits a 14-day internship when it answers one narrow question with one dataset and ends in one shareable deliverable. For BITS Pilani students, the project also has to sit around Practice School rather than compete with it, so the plan below works backwards from fixed days and cuts scope early instead of late.

Why a 14-day window forces a different kind of project

Most analytics project ideas students write down are really three-month ideas: “predict customer churn”, “build a sales dashboard for a retail chain”, “analyse Indian stock market behaviour”. Each hides a data-collection problem, a cleaning problem, a modelling problem and a communication problem. In 14 days, only one of those can be the centre of the project; the rest must be simplified or dropped.

The free two-week online internship from CareerSwami is built around a 14-day window, with a free offer letter and a free completion certificate after successful completion, across 26 tracks. That is short enough that a project’s scope decides whether you finish. The rest of this article shows how to size one properly.

Three scoping rules for a two-week analytics project

One question, written as a sentence with a verb. “Which product categories lost the most monthly revenue between two periods?” is scoped. “Sales analysis” is not.
One dataset you already have access to. If you need to scrape, request permission or merge five sources, the project is already a week behind. Public CSV files from government open-data portals or Kaggle-style repositories are the safest options; check each dataset’s licence before using it.
One deliverable a stranger can read in five minutes. A short notebook with a written summary, or a one-page dashboard with a findings note. Not both at full polish.

Free 2-Week CareerSwami Internship

Online, 14 days, with a free offer letter and a free completion certificate once you finish the work. Check with your college that it fits your requirement before you enrol.

Apply for the free internship

The 14-day milestone planner for a data analytics project

Assume roughly two to three focused hours per day, since many students will be handling coursework alongside. If you can give more, use the extra time for checking and writing, not for adding features.

Day	Milestone	Output you should have
1	Fix the question and success test	One-sentence question; a note on what answer would count as useful
2	Select and inspect dataset	Row/column count, data dictionary, licence note
3	Profile data quality	List of missing values, duplicates, odd types, outliers
4	Clean the data	Cleaned file plus a log of every change made
5	Exploratory pass	Five to eight basic charts or summary tables
6	Choose the core analysis	Short plan: method, variables, what you will not do
7	Mid-point review	Completed review template (below); scope decision recorded
8	Run the core analysis	First version of the key result
9	Sanity-check the result	Checks against a simple baseline or a hand calculation
10	Build the final visuals	Two to four clean charts with titles and labelled axes
11	Write the findings	Half-page summary: question, method, answer, limits
12	Assemble the deliverable	Notebook or dashboard in reading order
13	Peer or mentor read-through	List of fixes from one outside reader
14	Final fixes and submission	Final file, short README, submitted work

Notice that days 1 to 6 are about the data and not the model. Students who reach day 8 without a cleaned dataset usually end up rushing the part that employers read most closely: the findings.

Scope-cut matrix: what to drop first when time runs short

Decide cuts in advance, in order. Cutting under pressure on day 12 is how projects end up half-finished.

Scope element	Cut priority	What to do instead	What you lose
Second or third dataset	Cut first	Use one dataset and note the missing context as a limit	Breadth of insight
Predictive model	Cut second	Use descriptive statistics and segment comparisons	Forecasting claim
Interactive dashboard	Cut third	Static charts in a notebook or slide	Interactivity
Hyperparameter tuning	Cut fourth	Use a simple baseline and report it honestly	Marginal accuracy
Extra visualisations	Cut fifth	Keep only charts that support the answer	Decoration
Written findings	Never cut	Shorten if needed, but keep it	The whole point
Data cleaning log	Never cut	Keep a brief list of changes	Reproducibility

Mid-point review template for day 7

Spend 30 to 45 minutes on this at the end of day 7. Write answers down; do not just think them through.

Question check: Is my one-sentence question still the one I am answering? If it changed, what is the new wording?
Data readiness: Is the cleaned dataset final? List any known problems still open.
Evidence so far: What does the exploratory work already suggest? Write two sentences.
Remaining tasks: List every task left, with an hour estimate each. Add them up.
Time test: Does the total fit into the hours available over days 8 to 14, with at least one buffer day?
Cut decision: If not, which row from the scope-cut matrix will I apply now?
Help needed: What one question should I ask a mentor, peer or instructor before day 9?

Mid-point review (illustrative format)
Question: ...
Data status: cleaned / not cleaned
Evidence so far: ...
Tasks left (hours): ...
Total vs available: ...
Cut applied: ...
Question for reviewer: ...

Worked illustration: scoping a retail returns analysis

This example is illustrative. It describes a plan, not results from any run code or real dataset.

Initial idea: “Build a model to predict which online orders will be returned, plus a dashboard for the operations team.”

Problems at a glance: it needs order, product and customer data; it requires feature engineering; it needs a model and evaluation; and the dashboard adds another build step. That is four projects.

Rescoped question: “Which product categories have the highest return rate, and does the rate differ by order month?”

Plan:

Days 1 to 4: select one public e-commerce orders file, define “return” as a flag column, clean dates and categories.
Days 5 to 7: calculate return rate by category and by month; chart both.
Day 7 review: the total remaining estimate is 20 hours against about 18 available. Apply the scope-cut matrix: drop the predictive model and the interactive dashboard.
Days 8 to 10: test whether the category gap is large relative to the sample size, and add a simple comparison against the overall return rate.
Days 11 to 14: write findings, including that return reasons are not in the data, so causes cannot be claimed.

The finished deliverable is smaller than the first idea, but every statement in it can be defended. That is what a reviewer wants from a short internship.

Practice School at BITS Pilani and what it means for project timing

BITS Pilani students have an unusual constraint: Practice School is a formal, credit-bearing part of the degree, and the rules about it are strict. The Academic Regulations 2023 state that the normal duration of PS-I is about 8 weeks and PS-II about 22 weeks, with actual dates set each year. The same document says first-degree graduation requires either a Thesis or PS I plus PS II. Because this is a 2023 document, check that it is still the current version.

The regulations also say a student registering for any Practice School course cannot register simultaneously in any other course. That matters for planning: a short external project is realistic before or after a PS term, or alongside it only if it does not conflict with PS obligations. The regulations also give the Dean PS authority over terms and stations, so treat any PS timeline as decided by the institute, not by you.

The institute calendar PDF, in an excerpt, lists July 3 Practice School II registration and an October 28 mid-semester test for the 2026-27 first semester. Treat these as indicative and check the current calendar. A Dubai Campus document describes PS-I as eight weeks held in the summer after the second year; that is campus-specific and a 2023 excerpt.

Practically, a 14-day project works best in a gap: before PS reporting, between a semester and a station start, or during a break. Build your day 14 deadline so that it ends comfortably before any PS registration or reporting date. Also note that nothing found in the regulations text addresses online or remote internships, so do not assume a CareerSwami project counts toward PS; it is separate, personal portfolio work.

Steps to verify your own Practice School timing

Open the current Academic Regulations on the institute site and confirm whether the 2023 edition is still in force.
Download the current academic calendar and note your PS registration and reporting dates, then count back to pick a free 14-day window.
Email your PS unit or the Dean PS office, in writing, asking whether any external or online project is permitted during your PS term.
Ask your campus (Pilani, Goa, Hyderabad or Dubai) which PS schedule applies, since details can differ by campus.
Keep the reply, and record the dates you chose in your day 1 plan.

Choosing a dataset that fits a fortnight

Size: a file that opens in a spreadsheet or loads in seconds. Hundreds of thousands of rows are fine; many gigabytes are not.
Documentation: a data dictionary or clear column names. Guessing meanings costs days.
Time field: a date column makes trend questions possible without extra sources.
Licence: confirm you may reuse and publish analysis of it.
Relevance to your coursework: BITS Pilani’s MBA Business Analytics programme, for instance, lists topics such as Predictive Analytics, Time Series Analysis and Forecasting, and Data Visualization, Ethics and Data Privacy. If you study something similar, picking a dataset that echoes a course lets you reinforce what you are learning.

Writing the findings note so it survives a quick read

Use five short blocks: the question, the data (source, size, period), the method in two sentences, the answer with one or two numbers, and the limits. Limits are not weakness; they show you know what the data cannot say. A reviewer who sees “the data has no return reasons, so this does not explain why” trusts the rest more.

Daily habits that keep a short project on schedule

Begin each day by writing the one thing that must be done by the end of it.
Save a dated copy of your notebook or workbook each evening.
Log every cleaning change as you make it, not afterwards.
Stop adding charts once you have the ones that support the answer.
Reserve day 13 for another person’s eyes, even a classmate outside the field.

Frequently Asked Questions

Can a 14-day data analytics project really be portfolio-worthy?

Yes, if it is narrow and clearly explained. A focused question, a cleaned dataset, honest charts and a findings note show more skill than an ambitious model that was never finished.

Should I build a machine learning model in two weeks?

Only if the data is already clean and the question truly needs prediction. For most students, descriptive analysis with a clear comparison is safer and easier to defend. The scope-cut matrix lists the model as the second thing to drop.

How does BITS Pilani Practice School affect when I can do an outside project?

The 2023 regulations say a student registered for a Practice School course cannot register simultaneously in any other course. Plan any outside project for a gap in the calendar and ask the PS office in writing whether it is allowed during your term.

Will a CareerSwami internship count towards PS-I or PS-II?

No such arrangement is stated. CareerSwami offers a free offer letter and a free completion certificate after successful completion, and nothing more is claimed here. Whether anything counts toward PS is an institute decision, so ask the Dean PS office.

What if my dataset turns out to be too messy by day 4?

Switch datasets on day 4 at the latest, or narrow the question to the columns that are usable. Do not carry a dirty dataset past the mid-point review.

How many charts should the final deliverable have?

Two to four that each support the answer. Extra charts dilute the message and use time better spent on the findings note.

What should I do if I fall behind before day 7?

Use the mid-point review template to total the remaining hours, then apply cuts from the matrix in order. Cutting the second dataset and the model first protects the cleaning log and findings note.

Do the dates in the institute calendar apply to every campus?

Not necessarily. The institute runs several campuses, and the calendar facts here come from search excerpts. Confirm the dates for your own campus with the academic office before fixing your 14 days.