---

name: matlab-use-duckdb description: "Use DuckDB from MATLAB via Database Toolbox (R2026a+) as a non-math operations engine on large tabular files (CSV/Parquet/JSON) and as a zero-config embedded database. Use when connecting to DuckDB, querying CSV, Parquet, and JSON files directly with SQL, reducing or profiling large data before MATLAB analysis, creating portable development databases, or installing DuckDB extensions. Triggers on: DuckDB, duckdb(), large CSV/Parquet/JSON, file too large for readtable, filter/aggregate at source, deduplicate, reduce before analysis, profile large file, persistent file import, analytical engine, SQL on CSV, SQL on Parquet, SQL on JSON, query CSV with SQL, query Parquet with SQL, run SQL on files, SQL queries on files, query files directly, SQL without database, in-process SQL." license: https://www.mathworks.com/content/dam/mathworks/license/pmrl/license.md metadata: author: MathWorks version: "1.2"

MATLAB Database Toolbox Interface to DuckDB

Use when working with DuckDB databases from MATLAB using Database Toolbox. DuckDB is an embedded analytical database engine that ships with Database Toolbox starting in R2026a. It enables SQL-based analytics on files, out-of-memory data preprocessing, and portable development databases — all without external database server configuration.

When to Use This Skill

• Connecting to a DuckDB database (in-memory or file-based)
• Creating a new DuckDB database file for development workflows
• Querying CSV, Parquet, or JSON files directly with SQL
• Reducing or profiling large data before analysis
• Using DuckDB as a non-math operations engine (filter, aggregate, deduplicate, string cleanup, type cast, sort, sample)
• Installing and using DuckDB extensions
• Persisting file data into a .duckdb file for repeated queries
• User mentions keywords: DuckDB, duckdb(), analytical engine, embedded database, large CSV, large Parquet/JSON, file too large for readtable, filter/aggregate at source, deduplicate, reduce before analysis, profile large file, persistent file import, SQL on CSV, SQL on Parquet, SQL on JSON, query CSV with SQL, query Parquet with SQL, run SQL on files, SQL queries on files, query files directly, SQL without database, in-process SQL

When NOT to Use

• Single file that fits in memory — use native MATLAB I/O (readtable/readmatrix)
• Multi-file workflows — use datastore/tall first; DuckDB glob is the fallback, not the default
• Data exceeds memory and no SQL reduction will make it fit — use datastore/tall for streaming
• Math-heavy or numerically sensitive operations (normalize, windowed stats, outlier detection) — hand off to MATLAB after reduction
• Connecting to MySQL, PostgreSQL, SQLite, or other external databases — use their native interfaces or JDBC/ODBC
• Object-relational mapping — use ORM (ormread/ormwrite with Mappable classes)
• MongoDB, Cassandra, or Neo4j — use their dedicated Database Toolbox interfaces

On-Load Protocol

When this skill is loaded into a session where code already exists:

1. Audit the data pipeline — Identify how data enters the workflow:

   • Is data read via MATLAB I/O (readtable, readmatrix, readcell, readtimetable, parquetread, xlsread, csvread)?
   • Is data then written to DuckDB with sqlwrite before querying?
   • If yes: this is the load-then-query anti-pattern. DuckDB can likely read the source file directly via read_csv, read_parquet, or read_xlsx (excel extension).

2. Evaluate each data source against the decision framework:

   • Can DuckDB read this file type directly? (CSV, Parquet, JSON, Excel via extension)
   • Can filtering/aggregation be pushed into the SQL read?
   • Is the MATLAB I/O step a performance bottleneck?

3. Recommend architectural changes — Do not limit review to API correctness. Propose replacing readtable/xlsread + sqlwrite + query chains with fetch(conn, "SELECT ... FROM read_csv/read_parquet/read_xlsx(...)").

The highest-value patterns in this skill are architectural: file-analytics pushdown eliminates entire pipeline stages and can yield 10x+ speedups.

Pre-Flight Check

Before committing to the DuckDB path, estimate the file-size-to-available-RAM ratio (accounting for in-memory expansion of the file format) and route accordingly:

| Condition | Route | Rationale | |-----------|-------|-----------| | Single file, ratio < 0.5 | Native MATLAB I/O | Fits in memory; DuckDB adds unnecessary complexity | | Multi-file OR per-file processing | datastore/tall (fallback: DuckDB glob) | Streaming is the primary multi-file pattern | | Single file, ratio ≥ 0.5, SQL-expressible reduction | DuckDB: profile → operate → close | DuckDB is warranted | | Single file, ratio ≥ 0.5, no SQL reduction applies | datastore/tall | No reduction = no DuckDB advantage |

Operations Menu

DO in DuckDB SQL

| Operation | Example SQL | |-----------|-------------| | Filter rows | WHERE status = 'active' | | Aggregate | GROUP BY region, SUM(revenue), COUNT(*), AVG(price) | | Deduplicate | SELECT DISTINCT ... or ROW_NUMBER() OVER (PARTITION BY ...) | | String cleanup | TRIM(), LOWER(), REPLACE(), REGEXP_REPLACE() | | Type casting | TRY_CAST(col AS DATE), CAST(col AS DOUBLE) | | Sort | ORDER BY timestamp | | Sample | USING SAMPLE 10000 or TABLESAMPLE RESERVOIR(10%) | | Profile | COUNT(*), MIN/MAX, APPROX_COUNT_DISTINCT | | Limit | LIMIT 50000 |

DO NOT in DuckDB SQL

| Operation | Why Not | |-----------|---------| | Normalize / z-score | Requires population context | | Windowed statistics (moving avg, cumsum) | Fragile semantics, NULL handling differs | | Outlier detection | Statistical judgment call | | Interpolation / resampling | Domain-specific logic | | Signal processing | Not SQL's domain | | Machine learning features | Model-dependent |

SQL operations do not always match MATLAB semantics (NULL handling, sort order, deduplication). See reference/cards/reduce-large-data.md for known mismatches.

Reduction Workflow

If the pre-flight check routes to DuckDB, follow the profile → operate → close pattern in reference/cards/reduce-large-data.md.

DuckDB selects, reshapes, cleans, and reduces data. Once data fits in memory and the connection is closed, DuckDB's role is complete.

What Is DuckDB and Why Does Database Toolbox Ship It?

DuckDB is an embedded, serverless analytical database engine. Unlike MySQL or PostgreSQL, it requires no server, no configuration, and runs in-process within MATLAB.

Why it ships with Database Toolbox (R2026a+):

• Zero-config database — conn = duckdb() gives you a full SQL engine instantly.
• Analytical engine for files — Query CSV, Parquet, and JSON files directly with SQL without loading them into memory.
• Out-of-memory preprocessing — Filter, aggregate, join, and sort datasets larger than memory, then bring only results into MATLAB.
• Portable development databases — .duckdb or .db files work on any machine with Database Toolbox. No database setup needed.
• AI agent advantage — An agent's SQL knowledge directly translates to powerful analytical queries.

DuckDB does NOT replace MATLAB's file I/O (readtable, etc.) or datastore/tall. It is a performant alternative when data exceeds memory and the task reduces to a SQL-expressible operation. When no reduction applies, datastore/tall remains the correct path.

Critical Rules

Pre-Flight Gate

• ALWAYS run the pre-flight size-to-RAM check BEFORE writing any DuckDB code. If ratio < 0.5, STOP and use native MATLAB I/O instead. Do not proceed with DuckDB patterns.

Connection

• ALWAYS use duckdb() to connect — not database(), not JDBC, not ODBC.
• ALWAYS verify with isopen(conn) and close with close(conn).

API Surface

• All standard functions work: sqlread, fetch, execute, sqlwrite, sqlfind, sqlinnerjoin, sqlouterjoin, commit, rollback.
• DuckDB does NOT support databasePreparedStatement. Use execute or sqlwrite instead.
• Use ExcludeDuplicates via databaseImportOptions when reading from database tables (with sqlread). For direct file queries (read_csv/read_parquet via fetch), use SELECT DISTINCT in SQL.

Named Tables vs. File Queries

• ALWAYS use sqlread with RowFilter for simple row filtering on named database tables — not fetch with WHERE. Create a rowfilter object and pass it via the RowFilter name-value argument.
• Use fetch with SQL for aggregation, joins, or complex queries on named tables.
• ALWAYS use fetch (not sqlread) for file queries — they require SQL syntax like SELECT * FROM read_csv('file.csv').
• ALWAYS use single quotes for file paths inside SQL: read_csv('data.csv').

Connection Modes

| Goal | Connection | Why | |------|-----------|-----| | Analytical queries on files | duckdb() | No persistence needed; query files directly | | Temporary workspace | duckdb() | Fast, discarded on close | | Portable development database | duckdb("mydata.duckdb") | Creates a .duckdb or .db file; works on any machine | | Open existing database | duckdb("existing.db") | Read/write access to pre-existing .db or .duckdb file | | Read-only shared database | duckdb("shared.duckdb", ReadOnly=true) | Prevents accidental writes |

Common Patterns

Pattern 1: Analytical Engine on Files

conn = duckdb();
result = fetch(conn, "SELECT region, SUM(revenue) as total " + ...
    "FROM read_parquet('sales.parquet') " + ...
    "GROUP BY region ORDER BY total DESC");
close(conn);

Pattern 2: Out-of-Memory Preprocessing

conn = duckdb();
summary = fetch(conn, "SELECT date, AVG(value) as avg_val " + ...
    "FROM read_csv('huge_dataset.csv') " + ...
    "WHERE status = 'valid' " + ...
    "GROUP BY date ORDER BY date");
close(conn);
% summary fits in memory — continue with MATLAB analysis

Pattern 3: Read Named Table with RowFilter

conn = duckdb("inventory.duckdb", ReadOnly=true);
rf = rowfilter("quantity");
data = sqlread(conn, "products", RowFilter=rf.quantity < 10);
close(conn);

Pattern 4: Development Database

conn = duckdb("dev.duckdb");
sqlwrite(conn, "experiments", experimentData);
rf = rowfilter("score");
results = sqlread(conn, "experiments", RowFilter=rf.score > 0.8);
close(conn);

Pattern 5: Multi-File Query with Glob

Use when datastore/tall is unsuitable (e.g., SQL aggregation across files). For sequential per-file processing, prefer datastore.

conn = duckdb();
data = fetch(conn, "SELECT * FROM read_parquet('data/year=2024/*.parquet') " + ...
    "WHERE category = 'A'");
close(conn);

Pattern 6: Extensions

conn = duckdb();
execute(conn, "INSTALL httpfs");
execute(conn, "LOAD httpfs");
data = fetch(conn, "SELECT * FROM read_parquet('https://example.com/data.parquet') LIMIT 1000");
close(conn);

For Excel files, use the excel extension with read_xlsx (NOT st_read from spatial):

conn = duckdb();
execute(conn, "INSTALL excel");
execute(conn, "LOAD excel");
data = fetch(conn, "SELECT * FROM read_xlsx('report.xlsx')");
close(conn);

Pattern 7: Persistent Import from File

conn = duckdb("analytics.duckdb");
execute(conn, "CREATE TABLE events AS SELECT * FROM read_parquet('raw_events.parquet')");
% Future sessions: query by table name (no file re-read)
result = sqlread(conn, "events");
close(conn);

For detailed examples, see:

• File analytics and out-of-memory preprocessing: reference/cards/file-analytics.md
• Development database workflows: reference/cards/development-database.md
• DuckDB extensions: reference/cards/extensions.md

Common Mistakes

% WRONG — using database() or JDBC to connect to DuckDB
conn = database("", "", "", "org.duckdb.DuckDBDri