enterprise

Step-by-Step Guide to Learning MCP Servers for Data Analytics

• 7 min read

In Part 1, we learned what MCP servers are and why they matter for data analytics careers. Now let's get hands-on. This guide will walk you through setting up your first MCP server and using it to analyze real data.

What you'll accomplish:

Time required: 45-60 minutes

Experience needed: None (complete beginner friendly)

[← Read Part 1: How MCP Servers Are Changing Data Analytics Jobs](/mcp-servers-changing-data-analytics-jobs-beginners-guide/)

--

Software (All Free)

Hardware

--

Install VS Code

Install Node.js

Install Claude Desktop

--

Let's create realistic sales data to practice with.

Create a Project Folder

Create Sample Sales Data

In VS Code:

``csv

date,product,category,region,revenue,units_sold,customer_rating

2026-01-01,Widget A,Electronics,North,1500,30,4.5

2026-01-02,Widget B,Electronics,South,2300,46,4.2

2026-01-03,Widget A,Electronics,North,1800,36,4.5

2026-01-04,Gadget X,Home,North,3200,40,4.8

2026-01-05,Widget B,Electronics,East,1900,38,4.2

2026-01-06,Gadget Y,Home,South,2800,35,4.6

2026-01-07,Widget A,Electronics,West,1600,32,4.5

2026-01-08,Gadget X,Home,East,3400,42,4.8

2026-01-09,Tool Kit,Hardware,North,1200,24,4.0

2026-01-10,Widget B,Electronics,West,2100,42,4.2

2026-01-11,Gadget Y,Home,East,2900,36,4.6

2026-01-12,Tool Kit,Hardware,South,1100,22,4.0

2026-01-13,Widget A,Electronics,East,1750,35,4.5

2026-01-14,Gadget X,Home,West,3300,41,4.8

2026-01-15,Premium Tool,Hardware,North,4500,30,4.9

`

  • Save the file (Ctrl+S)

What this data represents:

  • Customer rating: 1-5 star rating

--

  • Step 3: Set Up Your MCP Server

Install the MCP Server Package

  • Type this command and press Enter:

`bash

npm install -g @modelcontextprotocol/server-filesystem

`

This installs the MCP server that lets AI read files on your computer.

Configure Claude Desktop

Add this configuration (replace YOUR_USERNAME with your actual computer username):

`json

{

"mcpServers": {

"filesystem": {

"command": "npx",

"args": [

"-y",

"@modelcontextprotocol/server-filesystem",

"/Users/YOUR_USERNAME/mcp-analytics-practice"

]

}

}

}

`

For Windows users, the path looks like:

`

"C:\\Users\\YOUR_USERNAME\\mcp-analytics-practice"

`

For Mac/Linux users, the path looks like:

`

"/home/YOUR_USERNAME/mcp-analytics-practice"

`

  • Restart Claude Desktop completely (quit and reopen)

Verify the Connection

When Claude Desktop restarts:

  • Click it to see available tools

--

  • Step 4: Analyze Your Data with AI

Your First Analysis

In Claude Desktop, type:

`

Please read the file sales_data.csv and tell me:

  • Which region performed best?

`

What happens:

  • You get answers in seconds

Expected Response

Claude should respond with something like:

"Let me analyze your sales data...

Total Sales for January: $33,150

Top Product by Revenue:

  • Gadget X: $9,900 (3 sales)

Best Performing Region:

  • North: $10,200 total revenue

Would you like me to dig deeper into any of these insights?"

--

  • Step 5: Ask Deeper Questions

Now try these follow-up questions:

Question 1: Category Performance

`

Compare the performance of Electronics vs Home vs Hardware categories. Which category has the best customer satisfaction? Which has the highest revenue per unit sold?

`

Question 2: Trend Analysis

`

Looking at the dates, is there any trend in sales over these 15 days? Are sales increasing, decreasing, or staying flat?

`

Question 3: Advanced Calculation

`

Calculate the average revenue per unit sold for each product. Then rank them from highest to lowest. Also tell me if there's a correlation between price per unit and customer rating.

`

--

Without MCP (Old Way)

To answer those questions manually, you would:

  • Create charts for trends

Time: 30-60 minutes

Skills needed: Excel formulas, pivot tables, statistics

With MCP (New Way)

You:

  • AI explained insights

Time: 2 minutes

Skills needed: Asking good questions

--

  • Step 7: Practice Exercises

Try these on your own:

Exercise 1: Filtered Analysis

`

Show me only the sales from the North region. What's the average customer rating for North region sales? How does it compare to other regions?

`

Exercise 2: Product Comparison

`

Compare Widget A and Widget B. Which one has better sales volume? Which has better revenue? Which has better customer ratings? Overall, which product is performing better?

`

Exercise 3: Hypothesis Testing

`

I think Gadget products have higher customer ratings than other categories. Test this hypothesis with the data and tell me if I'm right or wrong, with evidence.

``

--

Issue 1: "I don't see the hammer icon"

Solution:

Issue 2: "Claude says it can't access the file"

Solution:

Issue 3: "The analysis seems wrong"

Solution:

--

Technical Skills

Analytical Skills

Business Skills

--

Practice More

Learn Advanced Features

[→ Continue to Part 3: Advanced Data Analytics with MCP Servers](/advanced-data-analytics-mcp-servers/)

--

--

Documentation

Communities

Sample Datasets for Practice

--

Series: