---
title: timevector() | Tiger Data Docs
description: Create a timevector from time-value pairs for efficient time-series processing
---

Early access [1.3.0](https://github.com/timescale/timescaledb-toolkit/releases/tag/1.3.0)

Create a timevector object from time-value pairs. A timevector is a space-efficient intermediate representation that stores time-series data and can be passed to various analytic functions or used with pipeline operations.

## Samples

### Create a timevector from a table

Aggregate time-series data into a timevector:

```
SELECT timevector(time, weight) FROM samples;
```

### Store a timevector in a view

Create a view to store pre-aggregated timevectors:

```
CREATE VIEW series AS
    SELECT device_id, timevector(time, temperature) AS series
    FROM sensor_data
    GROUP BY device_id;
```

### Use timevector with downsampling

Pass a timevector directly to downsampling functions:

```
SELECT time, value::numeric(10,2)
FROM unnest((
    SELECT lttb(timevector(time, value), 20)
    FROM metrics
));
```

## Arguments

The syntax is:

```
timevector(
    time TIMESTAMPTZ,
    value DOUBLE PRECISION
) RETURNS Timevector
```

| Name  | Type             | Default | Required | Description                   |
| ----- | ---------------- | ------- | -------- | ----------------------------- |
| time  | TIMESTAMPTZ      | -       | ✔        | Timestamp for each data point |
| value | DOUBLE PRECISION | -       | ✔        | Value at each timestamp       |

## Returns

| Column     | Type       | Description                                                                                                                        |
| ---------- | ---------- | ---------------------------------------------------------------------------------------------------------------------------------- |
| timevector | Timevector | A timevector object containing the time-value pairs. Use with analytic functions, pipeline operations, or extract with `unnest()`. |