University of North Bengal Students: Analysing Tea and Tourism Data in a Short Internship

Yes, you can complete a credible tea or tourism analysis in a two-week internship, provided you keep the question narrow: one measure, one time period, one seasonal question. This article gives you a 14-day project template for time-series work, a checklist for judging your data, and a storyline for presenting findings to managers who do not read code.

Why tea and tourism data suit a short seasonality project

North Bengal’s two best-known economic stories, tea gardens and hill tourism, are both strongly seasonal. Tea production follows flushes and weather. Tourist arrivals follow holidays, festivals, monsoon closures and school calendars. That makes them ideal practice material: the pattern is real, the business question is obvious, and a manager can understand the answer without statistics training.

The free two-week online internship at CareerSwami (free offer letter, and a free completion certificate after successful completion, across 26 tracks) fits this scale well. Fourteen days is enough to clean a monthly series, describe its seasonality, test one simple forecast and present it. It is not enough for a full forecasting system, so scope accordingly.

Choosing one seasonality question you can finish in 14 days

A good question names a measure, a time grain and a decision. Weak: “Analyse tourism in Darjeeling.” Strong: “In which months does occupancy at a mid-sized hill-station property fall below its yearly average, and by how much?”

Sample questions by data type

• Tea production: Which months carry the largest share of annual output, and has that share shifted across the years in your data?
• Tea prices or auction averages: Do average prices rise or fall in predictable months once you remove the long-run trend?
• Tourist arrivals: How large is the gap between the peak and the lowest month, and is the gap widening?
• Hotel or homestay bookings: How far in advance do bookings arrive in peak versus lean months?
• Rainfall against either series: Does a dip in arrivals or output follow heavy-rain months, and with what lag?

Where to get tea and tourism figures, and how to judge them

Use only data you are allowed to use. Public government statistical releases, published industry reports and datasets that state their licence are safer than anything forwarded on messaging apps. If a figure cannot be traced to an originator and a date, treat it as unusable for a portfolio project.

The 14-day project template for a time-series and seasonality analysis

This is the core asset. Copy it into a document and fill in each row as you go.

Method notes that save time

• Start with the seasonal naive baseline. If a fancier model cannot beat it on a held-out year, say so honestly.
• Hold out the most recent full year for testing; never test on data used to fit.
• Use a multiplicative view when swings grow with the level, additive when they stay constant.
• Run the analysis twice if shock years exist: once including them, once excluded or flagged.
• Record every cleaning decision. A reviewer should be able to rebuild your table.

Worked illustration: monthly tourist arrivals at a hill destination

The numbers below are invented for illustration only. They do not describe any real destination, and no code was run to produce them.

Suppose you hold 4 years of monthly arrivals for a fictional hill town, averaging 10,000 a month in one year. You compute each month’s average across years and divide by the overall average to get a seasonal index.

The finding for a manager: “May runs at about 1.55 times a typical month, July at about 0.60. Staffing and promotions should follow that gap.” Then you test it: a seasonal naive forecast for the held-out year gives an error figure; a second method either beats it or does not. Report whichever happened.

A small starting sketch (illustrative, untested) for the index step in Python with pandas:

df["month"] = df["date"].dt.month
month_avg = df.groupby("month")["arrivals"].mean()
index = month_avg / df["arrivals"].mean()

A storyline outline for presenting seasonality to non-technical managers

Managers want a decision, not a decomposition. Build the deck around what changes on Monday.

1. The decision on the table. One sentence: “We need to plan staffing and stock for the lean months.”
2. The headline finding. “Demand peaks in May and bottoms out in July; the gap is about 2.5 to 1” (use your real numbers).
3. The picture. One clean chart: years overlaid by month, peak and lean labelled in plain words.
4. Is it getting better or worse? The trend, in one line, with the shock years explained.
5. What to expect next year. The forecast as a range, with how far past forecasts would have missed.
6. What we recommend. Two or three actions tied to specific months.
7. What we are unsure about. Data gaps, short history, outside events the data cannot predict.

Plain-language swaps

• “Seasonal index of 1.4” becomes “40% busier than a normal month”.
• “Residual variance” becomes “unusual swings we cannot explain by season”.
• “Holdout error” becomes “how wrong the forecast would have been last year”.

University of North Bengal students: fitting a data internship into the UG internship framework

The University’s FYUGP Regulations 2024-25 make internship compulsory for all students on the four-year programme, to be completed at the end of the 2nd or 4th semester during summer recess. The same document says the internship/apprenticeship component carries 2 credits. It also says internships are conducted, certified and marked by the respective affiliated colleges, with marks sent to the University, and that modalities are left to colleges.

In practice, this means your college, not the University centrally, decides what counts. The regulations as read do not specify required documents, an evaluation scheme or a minimum duration, and they do not address online internships. CareerSwami does not offer academic credit, so do not assume the certificate satisfies the 2-credit requirement. Treat the internship as skills-building and portfolio work unless your college confirms otherwise in writing.

Your data project may also connect to departments on campus. The Training & Placement Cell page lists a faculty committee that includes Tea Science and Economics, and says the cell offers placement assistance, counselling and workshops. That is a sensible place to ask for feedback on a tea or tourism project, though the page does not promise any particular outcome. Older regulations seen only as a search excerpt mention different internship lengths for a law programme, so check that you are reading the current version for your course.

Verification steps for NBU students planning a tea or tourism data internship

1. Ask your college’s internship coordinator or head of department in writing whether an online, self-arranged internship can be accepted for the 2-credit component.
2. Request the college’s internship format: required documents, minimum duration, report template and marking scheme.
3. Confirm which semester-end recess you are expected to use, and check the University’s Academic Calendar page for your programme’s current calendar.
4. Ask whether your report can use secondary public data on tea or tourism, and whether a faculty guide must approve the topic.
5. Download the current regulations yourself and check the clause for your programme before relying on any summary, including this one.
6. Ask the Training & Placement Cell whether it offers feedback sessions on student projects.

Common failure points in short seasonality projects, and fixes

Packaging the project for a portfolio

• A one-page summary leading with the finding, not the method.
• A notebook or script that runs from raw file to final chart.
• A data note listing sources, licences and cleaning decisions.
• A limitations paragraph: short history, missing months, unmodelled events.
• The offer letter and completion certificate from the internship, described accurately as free and online.

Frequently Asked Questions

How many years of data do I need for a seasonality analysis?

Aim for at least three full years of monthly data, ideally more. With fewer, a single unusual year can look like a season.

Which tool should I use: Excel, Python or Power BI?

Any of them can handle a monthly series. Excel suits seasonal indices and charts, Python suits decomposition and forecasting, and Power BI suits dashboards. Pick the one you can explain and reproduce.

Can I use tea or tourism data I find online?

Only if you can name the publisher and date and the reuse terms allow it. Avoid unattributed spreadsheets, and do not republish raw data without permission.

What do I do with pandemic or closure months?

Flag them, then run the analysis with and without those months. Tell the reader which result you recommend and why.

Will this internship count for the 2-credit internship in my NBU course?

That is decided by your college, since the regulations leave modalities to affiliated colleges. CareerSwami offers no academic credit, so ask your coordinator in writing before assuming anything.

Is a seasonal naive forecast too simple to present?

No. It is the standard yardstick. A model is only worth presenting if it beats that baseline on data it has not seen.

How do I explain a decomposition chart to a manager?

Skip the word “decomposition”. Show the yearly overlay, label the peak and lean months, and say the long-run direction in one sentence.

What should the final deliverable look like in 14 days?

A short deck of four to six slides, a reproducible notebook, a data-source note and a limitations paragraph.

Related Data Analytics guides

• Atria University Students: Making Your Analytics Internship Charts Readable and Honest
• Purnea University Students: Free Tools Roadmap for a Data Analytics Internship on a Budget
• PDM University Students: Writing the Final Data Analytics Internship Presentation
• All Data Analytics internship guides