Data Science Overview

Flow-Like brings powerful data science capabilities to a visual, no-code environment. Whether you’re exploring datasets, building ML models, or creating dashboards—you can do it all without writing code.

For Data Scientists

This section assumes familiarity with data science concepts like SQL, machine learning, and data visualization. We’ll show you how to apply your knowledge in Flow-Like’s visual environment.

What Can You Build?

Application Type	Description
Data Pipelines	Load, transform, and analyze data from multiple sources
Interactive Dashboards	Charts and visualizations with Nivo and Plotly
ML Workflows	Train and deploy classification, regression, and clustering models
Federated Analytics	Query across PostgreSQL, MySQL, Parquet, Delta Lake, and more
AI-Powered Analysis	Combine traditional ML with GenAI agents

Core Capabilities

1. Data Loading & Storage

Import data from CSVs, Excel files, databases, cloud storage, and APIs. Flow-Like’s storage system keeps your data organized and accessible.

👉 Learn about Data Loading & Storage

2. DataFusion SQL Analytics

Use SQL to query data from any source—local files, databases, or cloud data lakes. DataFusion unifies your data under a single query interface.

👉 Learn about DataFusion & SQL

3. Machine Learning Models

Train and deploy ML models for classification, regression, clustering, and dimensionality reduction using the linfa ML library.

👉 Learn about Machine Learning

4. Data Visualization

Create beautiful charts and dashboards using Nivo (17 chart types) and Plotly (scientific visualizations) directly in your A2UI interfaces.

👉 Learn about Data Visualization

5. GenAI for Data Science

Leverage AI agents for data analysis—natural language queries, automated insights, and intelligent data processing.

👉 Learn about AI-Powered Analysis

The Data Science Workflow

A typical data science workflow in Flow-Like:

┌──────────────────────────────────────────────────────────────────┐
│                                                                  │
│   1. LOAD DATA                                                   │
│      CSV, Excel, Parquet, APIs, Databases                        │
│                    │                                             │
│                    ▼                                             │
│   2. EXPLORE & TRANSFORM                                         │
│      DataFusion SQL, filtering, aggregation                      │
│                    │                                             │
│                    ▼                                             │
│   3. ANALYZE                                                     │
│      ├── Traditional ML (classification, clustering)            │
│      └── GenAI Agents (natural language analysis)               │
│                    │                                             │
│                    ▼                                             │
│   4. VISUALIZE                                                   │
│      Charts, dashboards, reports                                 │
│                    │                                             │
│                    ▼                                             │
│   5. DEPLOY                                                      │
│      Scheduled runs, APIs, Chat interfaces                       │
│                                                                  │
└──────────────────────────────────────────────────────────────────┘

Quick Example: Sales Analysis

Here’s what a sales analysis workflow might look like:

Read CSV ──▶ Mount to DataFusion ──▶ SQL Query ──▶ Bar Chart
    │              │                     │            │
    │              │                     │            │
    │         "sales_data"         "SELECT region,    │
    │                               SUM(revenue)      │
    │                               GROUP BY region"  │
    │                                                 │
    └─────────────────────────────────────────────────┘

1. Read CSV – Load your sales data file
2. Mount to DataFusion – Register as a SQL-queryable table
3. SQL Query – Aggregate by region
4. Bar Chart – Visualize results in A2UI

Supported Data Sources

Local Files

Format	Support
CSV	✅ Full (streaming, chunked reads)
Excel (.xlsx)	✅ Full (sheets, cells, tables)
Parquet	✅ Full (columnar, efficient)
JSON / NDJSON	✅ Full (with schema)

Databases

Database	Query	Write
PostgreSQL	✅	✅
MySQL	✅	✅
SQLite	✅	✅
DuckDB	✅	✅
ClickHouse	✅	✅
Oracle	✅	✅

Data Lakes

Format	Features
Delta Lake	Read, write, time travel
Apache Iceberg	Read, snapshots
Hive Partitioned	Parquet, JSON

Cloud Storage

Provider	Support
AWS S3	✅ Full
Azure Blob	✅ Full
Google Cloud Storage	✅ Full
AWS Athena	✅ Query

ML Algorithms Available

Category	Algorithms
Classification	Decision Trees, Naive Bayes, SVM
Regression	Linear Regression
Clustering	K-Means, DBSCAN
Dimensionality Reduction	PCA
Deep Learning	ONNX Runtime (YOLO, TIMM, custom models)

Visualization Options

Library	Chart Types
Nivo	Bar, Line, Pie, Radar, Heatmap, Scatter, Funnel, Treemap, Sunburst, Calendar, Sankey, Stream, Waffle, Chord + more
Plotly	Bar, Line, Scatter, Pie, Area, Histogram, Heatmap, Box, Violin

Prerequisites

Before starting with data science in Flow-Like:

1. Flow-Like Desktop installed (Download)
2. Data files or database connections ready
3. For ML: understanding of basic ML concepts
4. For AI analysis: API keys for LLM providers

Next Steps

Choose your starting point:

• Working with data? Start with Data Loading & Storage
• Need SQL analytics? Jump to DataFusion & SQL
• Building ML models? See Machine Learning
• Creating dashboards? Head to Data Visualization
• Want AI-powered insights? Explore AI-Powered Analysis