Unemployment Analysis — India

Problem

How hard did COVID-19 hit jobs in India? The raw dataset has monthly unemployment for 28 states and territories, split into rural and urban — too granular to answer the question at a glance, and easy to misread.

Approach

• Cleaning: found and dropped 28 fully-empty rows; parsed day-month-year dates explicitly.
• Correct aggregation: averaged ~50 state/area readings per month into one national figure, instead of plotting a misleading zig-zag of raw rows.
• Visualisation: trend lines before vs. during the pandemic, plus yearly and monthly box plots to show the spread across states.
• Plain-language insights summarised at the end of the notebook.

Results

	Result
Pre-COVID average (May 2019 – Mar 2020)	~9–10%, very stable
Peak national average	24.9% in May 2020 (23.6% in April)
Highest single reading	76.7% — urban Puducherry, April 2020
June 2020	Back down to ~12%, still above pre-pandemic levels

In April and May the spread between states exploded too — some regions were hit far harder than others.

What I learned

• Always check the data's granularity before plotting: grouping by date first fixed a chart that made the average look like 30–35%.
• Verify every claim with a number.
• Don't call something seasonal without enough history — 14 months shows a one-off shock, not a pattern.
• Cleaning comes first.

Dataset

"Unemployment in India" dataset (Kaggle).