AKTU Students: Data Analytics Internship Portfolio Guide for First-Time Applicants

A first-time data analytics applicant from AKTU needs three small, finished projects, each with visible SQL and a one-page insight summary. Three clean projects that a recruiter can open in two minutes beat ten half-built notebooks. This guide gives the blueprint, a SQL evidence checklist, a summary template and a worked illustrative example.

What a beginner analytics portfolio must prove

An internship reviewer who opens a fresher’s portfolio is asking four things. Can this person get messy data into shape? Can they answer a question with SQL? Can they separate a finding from a guess? Can they explain the result to someone who is not technical?

Your portfolio is evidence for those four questions. It is not a showcase of every tool you have touched. A beginner who shows one tool used carefully, with reasoning written down, looks stronger than one who lists fifteen tools with no outputs.

• Cleaning evidence: a before-and-after record of what was wrong in the data and what you changed.
• SQL evidence: saved queries that answer stated business questions, not just SELECT *.
• Reasoning evidence: a short note on what the numbers do and do not show.
• Communication evidence: a one-page summary that a manager could read in a minute.

The three-project blueprint

Each project is deliberately small, so it can be finished in three to five evenings. The three together cover the usual entry-level analytics tasks: descriptive reporting, comparison over time, and segmentation.

Choose datasets from open portals or public repositories whose licence permits reuse, and note the source and licence in each project README. Avoid any dataset that contains personal details of real people.

Project 1: Retail sales snapshot

Scope

Take an orders table, a products table and, if available, a customers or regions table. Load them into a local database such as SQLite, PostgreSQL or MySQL. The goal is a descriptive report on revenue, volume and top contributors.

Questions to answer in SQL

1. What is total revenue and order count per month?
2. Which five products produce the most revenue, and what share of the total do they hold?
3. Which regions have above-average order value?
4. How many orders have missing or zero values, and how were they handled?

Cleaning log

Keep a short table in the README listing each issue, how many rows it affected, and the decision you made. Typical issues are duplicate order IDs, inconsistent city spellings, dates stored as text and blank price fields. The decision matters as much as the fix: dropping rows and imputing values change results differently.

Project 2: Trend and seasonality check

Scope

Pick a dataset with at least 12 months of dated records. The goal is to describe movement over time and propose cautious explanations.

Questions to answer in SQL

1. What is the monthly total, and what is the month-over-month change in percentage?
2. What is the three-month moving average?
3. Which month is highest and lowest, and does the same pattern repeat across years if the data allows?
4. Does a weekday versus weekend split change the story?

This is where window functions earn their place in your portfolio. A query using LAG() to compute month-over-month change shows more skill than any course badge.

What to avoid

Do not claim a cause from a trend alone. Write “consistent with” or “may be related to” and list what extra data would test the idea, such as promotion calendars or price changes.

Project 3: Segmentation with a recommendation

Scope

Divide customers or users into three to five groups using simple, explainable rules. Recency, frequency and spend buckets work well and need no machine learning.

Questions to answer in SQL

1. How many customers fall in each segment, and what share of revenue does each produce?
2. How does average order value differ across segments?
3. Which segment shows the highest drop-off in repeat purchases?

End with one recommendation per segment and a note on how you would measure whether it worked. Reviewers look for this step because it separates an analyst from someone who only produces tables.

SQL evidence checklist

Screenshots of results are weak proof. A reviewer wants to read the query, see the output and understand why it was written. Use this checklist before publishing each project.

The one-page insight summary

Every project ends with a single page, saved as a PDF and linked from the README. Keep it to one page, with no code on it. Use this structure.

1. Headline (one sentence): the main finding with a number.
2. Question: what the analysis set out to answer.
3. Data: source, date range, row count, key exclusions.
4. Three findings: each with a figure and a plain-language reading.
5. One chart: the single visual that supports the headline.
6. Limits: what the data cannot tell you.
7. Recommended next step: one action or one extra dataset to request.

Worked example: data analytics internship portfolio (illustrative)

The following is an illustrative sketch, not a record of a real analysis. The table, column names and figures are invented to show the shape of the work; no query here has been run.

Suppose Project 2 uses a hypothetical orders table with order_date and amount. The monthly change query might look like this:

-- Q2: month-over-month revenue change
WITH monthly AS (
  SELECT
    DATE_TRUNC('month', order_date) AS month,
    SUM(amount) AS revenue
  FROM orders
  GROUP BY 1
)
SELECT
  month,
  revenue,
  LAG(revenue) OVER (ORDER BY month) AS prev_revenue,
  ROUND(
    100.0 * (revenue - LAG(revenue) OVER (ORDER BY month))
    / NULLIF(LAG(revenue) OVER (ORDER BY month), 0), 1
  ) AS pct_change
FROM monthly
ORDER BY month;

Now imagine the output showed a sharp rise in one month and a fall in the next. A weak summary would say “sales increased because of the festival”. A stronger illustrative summary would read:

• Headline: Revenue in the peak month was well above the surrounding months, followed by a fall the next month.
• Reading: The pattern is consistent with seasonal demand, but the dataset has no promotion or pricing field, so the cause cannot be confirmed.
• Limit: Only one year of data is available, so repeat seasonality is untested.
• Next step: Request promotion dates and compare against a second year.

Notice that the claim stays within what the data supports, and the next step is specific. That discipline is what reviewers remember.

Repository layout that is easy to review

Use one public repository per project, or one repository with three folders. Keep the structure identical across projects so a reviewer learns it once.

project-name/
  README.md          (question, data source and licence, findings, how to rerun)
  data/              (small sample or link; never personal data)
  sql/queries.sql    (numbered, commented, in run order)
  outputs/           (CSV results and chart images)
  summary.pdf        (the one-page insight summary)

The README should open with the headline finding, not with installation steps. A busy reviewer may read only the first ten lines.

Portfolio scorecard: check yourself before applying

If any row sits in the weak column, fix it before applying. Two projects at the solid level are better than three with gaps.

AKTU and the data analytics portfolio: where the internship rule fits

AKTU is an affiliating university set up by the Uttar Pradesh government, and its admission portal describes jurisdiction over the whole state and 700+ affiliated colleges, both government and private. For a portfolio builder, this means the university publishes the scheme, but day-to-day arrangements (who guides you, what counts as a report) are likely to be handled by your own college and department.

The most relevant rule for a first internship is in the B.Tech 2nd-year scheme. Excerpts of the 2023-24 schemes, such as the Textile Technology scheme, state that a 4-week internship or mini project is done in the summer break after semester 4 and assessed in semester 5. The same excerpt lists the paper BCC351 at 100 marks and 2 credits. A Plastic Engineering scheme carries the same wording. A third-party copy of the CSE-group scheme (a mirror, not an official page) repeats it for Computer Science, IT and AI branches. These are excerpts from one scheme year, so the current version for your batch must be checked.

In practice, a 4-week window fits the 14-day first-project schedule in this guide: finish the retail sales project early, then use the remaining weeks for the trend and segmentation projects. Note that the older 2019-20 scheme, seen in a third-party copy, placed this after semester 2, so advice from seniors on an older scheme may not match yours.

Whether an online data analytics internship can satisfy this credit-bearing requirement is not established by the sources reviewed, so it should be treated as a question for your college. For placement-side contact, the UP government’s URISE page for AKTU’s Central Training & Placement Cell lists industry announcements, including a virtual internship offer for B.Tech, M.Tech, MBA and MCA students dated 2025. Check whether any such offer is current before relying on it.

Questions to settle before you count this work towards BCC351

• Find the scheme PDF for your admission year and branch on the official AKTU site and read the internship or mini project clause yourself.
• Ask your department coordinator whether a self-directed analytics portfolio or online internship is accepted as the 4-week requirement.
• Ask what documents your college wants (report format, completion letter, faculty mentor) and the submission date before semester 5.
• Confirm your semester 4 break dates from your college’s current academic calendar.
• Ask the training and placement office whether any current industry or virtual internship programme is open to your branch.

Where an internship fits in

If you want a structured short deadline to build the first project against, the free two-week online internship at CareerSwami offers 26 tracks across fields, a free offer letter and a free completion certificate after successful completion. It runs for 14 days, which is enough to finish one portfolio project properly. Treat the certificate as a supporting line on your resume, and let the project itself carry the weight.

Questions to settle before you count this work towards BCC351

These points vary by affiliated college, so settle them in writing before you plan around them.

• Ask your department coordinator or training and placement cell whether an online internship can be used for any internship or training requirement in your current scheme. Request the answer by email.
• Find your branch’s current syllabus and evaluation scheme on the university website and ask your faculty advisor which semester carries any project or internship component.
• Ask whether your college requires prior approval, a specific letter format or a faculty mentor for an external internship.
• Ask what documents are accepted as proof of completion, such as an offer letter, certificate or a supervisor’s report.
• Check whether lab machines or institute email accounts can be used for the work, or whether personal devices are expected.

A 14-day schedule for the first project

Repeat the cycle for Projects 2 and 3, with each cycle faster than the last.

Presenting the portfolio in an application

• Put a single link at the top of your resume, pointing to a page that lists all three projects.
• Write one resume line per project that begins with the finding, for example “Identified the top five products contributing the majority of revenue using SQL aggregation”, and only if your own result supports it.
• List tools after the results, not before.
• In a cover note, mention one limitation you found in your own work. It signals analytical honesty.
• Make sure every link works while logged out.

Common mistakes first-time applicants make

• Copying a tutorial exactly. Reviewers have seen the same famous datasets many times. Change the questions, or add your own angle, so the work is demonstrably yours.
• Showing charts without numbers. Every chart needs a title that states the takeaway.
• Skipping validation. If your totals do not match the raw file, nothing else can be trusted.
• Overfitting to tools. A basic spreadsheet chart with a sound argument beats an elaborate dashboard with none.
• Uploading sensitive data. Never publish personal, confidential or college-internal records.

Frequently Asked Questions

How many projects do I need for a first analytics internship application?

Three small, finished projects are enough. Two strong ones are acceptable if each has SQL, a cleaning log and a one-page summary. Quality of reasoning matters more than count.

Do I need Python or is SQL enough?

For a first application, SQL plus a spreadsheet or a BI tool is enough to show the core skills. Python helps for cleaning and charts, but it is not a substitute for readable SQL and clear findings.

Which database should an AKTU student use for SQL portfolio projects?

SQLite is the easiest to install and sufficient for these projects. PostgreSQL or MySQL is a good choice if you want closer-to-industry practice, especially for date functions and window functions.

Can I use a dataset from my college coursework?

Only if you are allowed to share it and it contains no personal or confidential records. When in doubt, use an open public dataset and keep the coursework version private.

Is a certificate enough without a portfolio?

A certificate shows participation. A portfolio shows what you can actually do. Reviewers for analytics roles usually look for the second, so treat any certificate as supporting evidence.

Will an online internship count for my AKTU requirements?

That depends on your college and the rules for your scheme and branch. Ask your department coordinator or training and placement cell for a written answer rather than relying on a general assumption.

How long should the one-page summary be?

One page, about 250 to 350 words plus one chart. If it needs more space, the headline finding is probably not yet clear.

What if my results are unremarkable?

That is fine. A clearly explained “no strong difference between segments” finding, with the evidence and the limits stated, shows better judgement than a forced conclusion.

Related Data Analytics guides

• MAKAUT Students: How to Write and Structure a Data Analytics Internship Report
• GTU Students: Choosing a Data Analytics Internship Project and Preparing the Skills for It
• Comparing Data Analytics Internship Roles for Savitribai Phule Pune University Students
• All Data Analytics internship guides