Not a Single Person Passed the Interview.

Picture this: a room full of aspiring data analysts, all eager to impress. They were handed a popular dataset—the Madhav Store Dataset—to analyze. The result? ?????? ?? ???????????? ???????????? ????????????.

Why did they all fail?

?

I was called in to assess their performance. These were smart, hardworking individuals, but they missed key details that could have set them apart.

?

Let me break down ?? ?????? ?????????????? ???????? ????????’?? ???????? —lessons every analyst should learn:

??.??????? ?????????????????????????? ???????????????? Most of them jumped straight into profitability analysis, using only COGS (Cost of Goods Sold). But profitability isn’t that simple. To analyze profits correctly, you must also consider:

1. Operating Expenses (OPEX): This includes everything from salaries to utility costs.
2. Taxes: Ignoring tax implications gives an incomplete picture.
3. Depreciation & Amortization: These are crucial for understanding long-term asset costs.
4. Interest Payments: If a company has debt, interest significantly impacts profits.

When these factors are excluded, any "profitability" analysis is incomplete and misleading.

?

??.??????? ?????????????????? ???????????????????? They tried to fit all KPIs—orders, profits, sales, and quantities—into a single dashboard. This doesn’t work because it creates confusion. Here’s the better approach:

1. Focus on one key metric per dashboard. For example, a dashboard for revenue or one for quantity.
2. If you must show multiple KPIs, create separate dashboards for each. This ensures clarity and helps tell a more compelling story.

?

??.??????? ???????????????????? ?????? ???????? Many reported that the number of orders was 1,500. The correct number? 500 orders. How did this happen? They misunderstood how the dataset was structured. This is a reminder that attention to detail is everything in data analysis.

?

??.??????? ???????? ?????????? ?????????????????????????? Some analysts merged the Orders table and the Details table into one. While this might seem easier at a beginner level, it’s inefficient for advanced analysis. Why?

1. Merging the tables increases the size of the dataset. For instance, instead of refreshing 7 columns from the Details table, you’re refreshing all 12 columns from the merged table. This slows performance.
2. The proper method is to create a relationship between the tables. This ensures faster refresh rates and better optimization.

?

??.??????? ???????????????? ?????????????? None of them considered the population of India or the characteristics of different states. Why does this matter?

1. A state with a higher population is naturally likely to have more sales. Ignoring this gives a skewed analysis.
2. India has 28 states and 8 Union Territories. Some states are larger, wealthier, or more urbanized than others. These factors play a significant role in sales patterns.

For example, a state with a population of 5,000 may not have the same sales as a state with 500,000, that’s a key insight worth exploring. Ignoring context leads to missed opportunities for deeper analysis.

These are common mistakes, but they’re entirely avoidable with the right training. If you’ve worked on this dataset before, I recommend revisiting it and making these corrections. If you haven’t, I suggest you start now.

?

If you want to practice, grab the Madhav Store Dataset here: (https://www.kaggle.com/datasets/saadharoon27/madhav-store-dataset?resource=download )

?

?????? ??????????????????????: I will be holding a free class on how to clean and transform this dataset. If you're interested, use this link to fill out the form: https://forms.gle/jMnoqqyB8HkhFUdo7

?

Coach Anietie