Singdata Lakehouse MCP Server User Case

End-to-End Natural Language Conversational Approach from Data Acquisition to Visual Analytics

Can you help me create a volume? Name it NYC Green Taxi Data
The bucket is mcp-data-hangzhou, path is /nyc-green-taxi/
Download each month's data from 2025 from the NYC website into this volume
Please analyze it and generate a dashboard

Case Overview

This case demonstrates how to use Claude Desktop with Singdata Lakehouse MCP Server to complete an end-to-end data science workflow from data acquisition, storage, and analysis to visualization through natural language conversation.

Case Scenario: Analyze New York City Green Taxi 2025 operational data and build a business insights dashboard

Technologies Involved:

  • Claude Desktop (AI assistant interface)
  • Singdata Lakehouse (cloud-native data lakehouse)
  • MCP Server (Model Context Protocol service)
  • NYC TLC public dataset

Solution Highlights

Zero-Code Data Analysis

  • Describe requirements in natural language, and AI automatically executes all complex data operations
  • No need to write SQL, Python, or any other code
  • Automatically handles data download, cleaning, storage, and analysis

Enterprise-Grade Data Infrastructure

  • Cloud-native Lakehouse architecture supporting petabyte-scale data processing
  • Automatic backup and version management
  • High availability and security guarantees

Intelligent Data Insights

  • AI-driven data analysis and pattern discovery
  • Automatically generates business insights and recommendations
  • Interactive visualization dashboard

Implementation Steps

Step 1: Environment Preparation

Prerequisites:
āœ… Claude Desktop client
āœ… Singdata Lakehouse account
āœ… MCP Server configuration

Step 2: Data Acquisition and Storage

User Input:

AI Execution Process:

  1. Create Storage Space

    • Automatically create external Volume: nyc_green_taxi_data
    • Connect OSS storage: oss://mcp-data-hangzhou/nyc-green-taxi/
    • Configure auto-refresh and recursive scanning

  2. Batch Data Download

    • Intelligently identify NYC TLC official data source
    • Download Jan-Jul 2025 Parquet files
    • Automatically handle file format and compression

  3. Data Ingestion

    • Create unified data table: nyc_green_taxi_2025
    • Standardize data schema and field types
    • Add data month identifier field

Results Summary:

  • Total records: 351,612 trips
  • Time span: Jan-Jul 2025
  • Data size: approx. 8.5 MB (Parquet compressed)
  • Processing time: <5 minutes

Step 3: Intelligent Data Analysis

User Input:

AI Analysis Dimensions:

  1. Time Trend Analysis

    • Monthly trip volume and revenue changes
    • 24-hour usage patterns
    • Weekday vs. weekend comparison

  2. Spatial Distribution Analysis

    • Identify popular pickup locations
    • Discover high-value areas
    • Trip distance statistics

  3. Business Metrics Analysis

    • Fare distribution and structure
    • Payment method preferences
    • Congestion fee impact assessment

  4. User Behavior Analysis

    • Passenger count patterns
    • Trip duration distribution
    • Tipping habits !

Key Findings

Revenue Trends

  • Peak month: May revenue of $1.33 million, up 31.6% month-over-month
  • Average fare: $24.45, showing an upward trend
  • Total revenue: $8.6 million accumulated over 7 months

Time Patterns

  • Peak hours: 4-6 PM, over 25,000 trips per hour
  • Nighttime premium: Average fare of $30+ from 2-4 AM
  • Weekend effect: Weekend fares are 3.9% higher than weekdays

Geographic Distribution

  • Popular areas: Locations 74 and 75 contribute 38.7% of total trip volume
  • High-value points: Location 244 averages $35.47 per fare
  • Coverage: Primarily serves outer boroughs and Upper Manhattan

Payment Trends

  • Digital adoption rate: 74% use credit card payments
  • Cash share: 26%, showing a declining trend
  • Congestion fee: 9.6% of trips include it, averaging $0.75

Visualization Dashboard

Interface Features

  • Responsive design: Adapts to desktop, tablet, and mobile
  • Interactive charts: Supports zoom, filter, and drill-down
  • Real-time data: Configurable auto-refresh
  • Modern UI: Gradient backgrounds, animated effects

Chart Types

  • šŸ“ˆ Trend charts: Dual-axis display of monthly revenue and trip volume
  • šŸ“Š Bar charts: 24-hour usage patterns, popular location rankings
  • šŸ© Donut charts: Fare distribution, payment method breakdown
  • šŸ“‹ Metric cards: Key KPIs at a glance

Business Value

šŸš€ Efficiency Gains

  • Development time: Reduced from weeks to minutes
  • Technical barrier: Zero code, business users can directly use it
  • Iteration speed: Rapid response to changing requirements

Insight Quality

  • Multi-dimensional analysis: Full coverage of time, space, and user behavior
  • Intelligent discovery: AI identifies patterns easily overlooked by humans
  • Decision support: Direct business recommendations and predictions

Cost Optimization

  • Infrastructure: Cloud-native architecture, pay-as-you-go
  • Human resources: Reduced dependency on specialized data teams
  • Maintenance cost: Automated operations, lower management overhead

Applicable Scenarios

šŸ“Š Business Analytics

  • Sales data analysis and forecasting
  • Customer behavior insights
  • Market trend research
  • Competitive analysis

Operations Optimization

  • Supply chain efficiency analysis
  • Equipment maintenance prediction
  • Quality control monitoring
  • Cost structure optimization

Product Analytics

  • User retention analysis
  • Feature usage statistics
  • A/B test evaluation
  • Product roadmap planning

Marketing Analytics

  • Advertising effectiveness evaluation
  • Channel ROI analysis
  • User profiling
  • Precision marketing strategies

Core Tools Deep Dive

1. Claude Desktop

Role: Intelligent conversational interface and AI assistant

Core Capabilities:

  • šŸ¤– Natural language understanding: Converts user requirements into specific operational commands
  • šŸ”„ Multi-modal interaction: Supports text, charts, code, and other output formats
  • šŸ§  Intelligent reasoning: Performs logical reasoning and decision-making based on context
  • šŸ“‹ Task orchestration: Automatically plans and executes complex data workflows

Application in this case:

  • Understands natural language commands such as "create a volume" and "download data"
  • Automatically plans the complete workflow from data acquisition to visualization
  • Intelligently analyzes data patterns and generates business insights
  • Creates interactive HTML dashboards

2. MCP Server (Model-Context Protocol)

Role: Bridge between AI and external systems

Core Capabilities:

  • šŸ”Œ Protocol translation: Converts AI commands into database operations
  • šŸ›”ļø Security proxy: Provides secure data access control
  • āš” Efficient communication: Optimizes interaction performance between AI and data systems
  • šŸ”§ Tool integration: Supports 70+ specialized data operation tools

Key MCP Tools:

šŸ“Š Data Management Tools

āœ“ create_volume - Create data storage volumes
āœ“ import_data_src - Import data from URLs
āœ“ preview_volume_data - Data preview and validation
āœ“ list_files_on_volume - File management

šŸ—ļø Table Structure and Volume Tools

āœ“ create_table - Create data tables
āœ“ desc_object - View object details
āœ“ show_object_list - List database objects
āœ“ alter_table - Modify table structure
āœ“ create_volume - Create data lake storage volumes

šŸ” Query and Analysis Tools

āœ“ read_query - Execute SELECT queries
āœ“ write_query - Execute write operations
āœ“ vector_search - Vector similarity search
āœ“ match_all - Full-text search

šŸŒ Data Acquisition Tools

āœ“ smart_crawl_to_volume - Intelligent web data crawling
āœ“ crawl_single_page - Single page content extraction
āœ“ import_data_from_db - Database data import

šŸ”— Connection Management Tools

āœ“ create_storage_connection - Storage connection configuration
āœ“ create_api_connection - API service connection
āœ“ switch_context - Context environment switching

3. Singdata Lakehouse

Role: Cloud-native data lakehouse integrated platform

Core Architecture:

  • šŸ—ļø Storage-compute separation: Independent scaling, cost optimization
  • šŸ“Š Multi-mode support: Batch processing, stream processing, interactive queries
  • āš” Query acceleration: Intelligent caching and pre-computation

Data Storage Layer:

  • Volume management: Unified data storage abstraction
  • Multi-cloud support: OSS, S3, COS, and other object storage
  • Format compatibility: Parquet, ORC, JSON, CSV, and more
  • Version control: Data lineage and time travel

Compute Engine:

  • SQL engine: Standard SQL, compatible with multiple dialects
  • Vector computing: Supports AI/ML workloads
  • Stream processing: Real-time data pipelines
  • Federated queries: Unified queries across data sources

Application in this case:

  • Created nyc_green_taxi_data volume for storing raw data
  • Established nyc_green_taxi_2025 table for structured storage
  • Executed complex multi-dimensional analysis queries
  • Supported sub-second query response for 350,000+ records

4. External Data Sources

NYC TLC (Taxi & Limousine Commission)

Data Characteristics:

  • šŸ“… Regular updates: Monthly releases, 2-month latency
  • šŸ“Š Data quality: Officially authoritative, standardized format
  • šŸ’¾ Storage format: Parquet compressed format
  • šŸŒ Access method: CDN accelerated download

Data Fields (21 core fields):

ā€¢ vendorid - Service provider ID
ā€¢ lpep_pickup_datetime - Pickup time
ā€¢ lpep_dropoff_datetime - Dropoff time
ā€¢ pulocationid/dolocationid - Pickup/Dropoff location
ā€¢ trip_distance - Trip distance
ā€¢ fare_amount - Base fare
ā€¢ total_amount - Total fare
ā€¢ payment_type - Payment method
ā€¢ cbd_congestion_fee - Congestion fee (new in 2025)

5. Visualization Component (Claude Desktop Artifacts)

Chart.js + Custom HTML

Chart Types:

  • šŸ“ˆ Dual-axis line charts: Monthly trends (trip volume + revenue)
  • šŸ“Š Bar charts: 24-hour patterns, popular locations
  • šŸ© Donut charts: Fare distribution, payment methods
  • šŸ“‹ Metric cards: Key KPI display

Interactive Features:

  • āœØ Hover effects: Display detailed values
  • šŸ”„ Responsive: Adapts to different screen sizes
  • šŸŽØ Animations: Smooth transition effects
  • šŸ“± Mobile-friendly: Touch optimized

Technical Architecture

Data Flow Architecture

Tool Call Examples

Actual tool call sequence in this case:

1. get_current_context() - Retrieve current environment information
2. create_volume() - Create nyc_green_taxi_data volume
3. import_data_src() Ɨ 7 - Batch download 7 months of data files
4. create_table() - Create nyc_green_taxi_2025 unified table
5. write_query() Ɨ 7 - Insert each month's data into the main table
6. read_query() Ɨ 8 - Execute multi-dimensional analysis queries
7. add_data_insight() - Save analysis insights
8. artifacts.create() - Generate visualization dashboard

Performance Optimization Features

  • Query cache: Millisecond-level response for repeated queries
  • Intelligent indexing: Automatic index creation for frequently queried fields
  • Partitioning strategy: Monthly partitioning to improve query efficiency
  • Compressed storage: Parquet format saves 90% storage space
  • Parallel processing: Concurrent multi-file download and processing

Quick Start

1. Apply for Trial

  • Visit the Singdata official website to apply for an account
  • Download and install the Claude Desktop client
  • Configure the MCP Server connection

2. Prepare Data

  • Define analysis goals and data sources
  • Prepare data access permissions
  • Plan storage structure

3. Start the Conversation

Example dialogue:
User: "I want to analyze our company's sales data"
AI: "Sure, let me help you with that analysis..."

4. Obtain Insights

  • Review automatically generated analysis reports
  • Engage in conversational deep dives with AI
  • Export results and visualizations

Summary

Singdata Lakehouse MCP Server, combined with Claude Desktop, provides enterprises with an unprecedented data analysis experience:

Core Advantages

  • Natural interaction: Complete complex data analysis through conversation
  • Zero technical barrier: Business users directly obtain insights
  • Enterprise-grade capability: Supports large-scale, high-concurrency scenarios
  • Rapid iteration: From idea to insight in minutes

Innovative Breakthroughs

  • Redefining human-machine collaboration models
  • Dramatically lowering the barrier to data analysis
  • Accelerating enterprise digital transformation
  • Unlocking the true value of data

Future Outlook

As AI technology continues to advance, this "conversational data analysis" approach will become an enterprise standard, enabling every business user to become a data analyst and letting data truly drive every business decision.

Start your intelligent data analysis journey today!

N8N Integrated with Lakehouse MCP Server
NYC Green Taxi Data Clean Transform with MCP Server
Singdata Ā© 2024 Singdata, Inc. All Rights Reserved.