# Creating a reduced SPI set

## Filtering an SPI Set

If you would like to filter SPIs based on their corresponding keywords, you can do so using the **`filter_spis`** function in *pyspi*. This will return a new config file containing the reduced SPI set that you can provide to the `Calculator` object. 

As an example, let's use the filter to obtain a subset of SPIs with the keywords `linear` and `signed`:

```python
from pyspi.utils import filter_spis

# let's only select SPIs with the keywords "linear" and "signed"
# we will need to provide these keywords in a list
keywords = ["linear", "signed"]

# we will need to specify the location of the original config.yaml
filter_spis(keywords=keywords, output_name="linear_signed.yaml")
```

*Note* that here we do not specify a source config.yaml file. By **default**, the `filter_spis` function will use the **original config.yaml** (containing the full set of SPIs) located in the *pyspi* directory.  If you would like to provide a custom YAML as the source file, you can provide the location of the file to the function with the optional keyword `configfile.`

<details>

<summary>[OPTIONAL] Filtering from a user-specified YAML</summary>

Here, we tell the filter function to use a custom yaml file `myconfig.yaml` located in the current working directory:

```python
filter_spis(keywords=keywords, output_name="linear_signed.yaml", configfile="./myconfig.yaml") 
```

You must ensure that the custom config YAML conforms to the structure of a *pyspi* YAML otherwise the filter function will not work as expected. See [manually specifying a reduced set ](#manually-specifying-a-reduced-set)for more information about the structure *pyspi* expects.

</details>

We should obtain a reduced subset of SPIs with the keywords "linear" and "signed". Let's now compute this reduced set on some data:

```python
from pyspi.calculator import Calculator
import numpy as np

data = np.random.randn(3, 100) # your time series data
calc = Calculator(configfile="linear_signed.yaml", dataset=data)

# print the number of SPIs initialised in the Calculator instance
print(calc.n_spis)
```

***

## Manually specifying a reduced set

You can use a subset of the SPIs by copying a version of the <mark style="color:red;">**`config.yaml`**</mark> file to a local directory and manually removing those you don’t want the calculator to compute. First, copy the <mark style="color:red;">**`config.yaml`**</mark> file to your workspace:

```python
import os, shutil
import pyspi

# chose the location where you would like to save the copy 
destination = "./myconfig.yaml"
# get the pyspi directory where config.yaml is located
source = os.path.dirname(pyspi.__file__) + "/config.yaml"
shutil.copy(source, destination)
```

Once you've got a local copy of <mark style="color:red;">**`config.yaml`**</mark> edit the file to remove any SPIs you're not interested in. A minimal configuration file might look like the following if you're only interested in computing a covariance matrix using the *maximum likelihood estimator*:

{% code title="myconfig.yaml" %}

```yaml
# Basic statistics
.statistics.basic:
    # Covariance
    covariance:
       labels:
          - undirected
          - linear
          - signed
          - multivariate
          - contemporaneous
       dependencies:
       configs:
        # Maximum likehood estimator
        - estimator: EmpiricalCovariance
```

{% endcode %}

Ensure you retain the `labels` and `dependencies` keys from the original <mark style="color:red;">**config.yaml**</mark> file when creating your custom set.&#x20;

When you instantiate the calculator, instead of using the default <mark style="color:red;">**`config.yaml,`**</mark> you can input your bespoke configuration file:

```python
from pyspi.calculator import Calculator

calc = Caculator(dataset=dataset, configfile='myconfig.yaml')
```

Then use the calculator as normal (see [Getting Started](/pyspi/installing-and-using-pyspi/usage/walkthrough-tutorials/getting-started-a-simple-demonstration.md)).&#x20;


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://time-series-features.gitbook.io/pyspi/installing-and-using-pyspi/usage/advanced-usage/creating-a-reduced-spi-set.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.