---
title: Filtering Columns
slug: filtering-columns
icon: 📖
docTags: 
createdAt: 2025-04-10T20:26:30.543Z
---

***Organizing your Insights***

## Filtering and Sorting Columns 

As you create data columns, you might want to narrow your results to specific areas of interest. This can be completed by **Filtering** and/or **Sorting** results.

### Filtering

::::VerticalSplit{layout="middle"}
:::VerticalSplitItem
Filtering is a powerful tool for insight generation in Tabular. Select the **filter icon **on any column header to bring up the menu:

Here, you can search and select exactly the result(s) needed for your project goals.

Note, you can **simultaneously filter across  multiple columns** to perform complex data manipulation. 
:::

:::VerticalSplitItem
![](https://api.archbee.com/api/optimize/A8WZzaoFAHNpve3k5PHtC/ma66vy1CIvvPel3DEJdp1_.blob)
:::
::::

**Filters also allow users to specify AND / OR criteria for multi-keyword matching.**

::::VerticalSplit{layout="middle"}
:::VerticalSplitItem
For example, if I am looking to limit a literature search to **studies performed in the China OR United States**, I can create the following filter:

Users can specify up to 4 keywords per filter.
:::

:::VerticalSplitItem
![](https://api.archbee.com/api/optimize/A8WZzaoFAHNpve3k5PHtC/Bl2y8eQcTdQoIW90M4yjx_image.png)
:::
::::

**Filtering via Yes / No Columns**

An effective way to filter results in Tabular is to create precise custom columns leveraging Reliant’s powerful **Yes / No output type**. These columns are fine tuned to think carefully about the question, and return a binary *Yes* / *No* answer whenever supported by evidence. When no supporting evidence, or conflicting evidence is found, the model will return *Unclear*.

Say I run a PubMed search on: ***Clinical Trial Papers in Huntington’s disease in the last 5 years***.  I am interested in quickly classifying clinical trials that leverage viral vector delivery mechanisms. To quickly identify and filter my results, I can use a Yes / No column: 

::::VerticalSplit{layout="middle"}
:::VerticalSplitItem
**Yes / No Column Prompt:** *Does this clinical trial discuss the use of a viral vector (ex. AAV, LV, etc.) for delivery of the primary intervention?*
:::

:::VerticalSplitItem
![](https://app.archbee.com/api/optimize/A8WZzaoFAHNpve3k5PHtC/0lcqIE3AADRqeS3syRUbZ_.blob)
:::
::::



### Sorting

::::VerticalSplit{layout="middle"}
:::VerticalSplitItem
Select the carrot icon next to any column header to **sort by alphabetical order. **

When dealing with long text columns, toggle the **Wrap Column** feature to clean up your table view. 

This menu will also enable the user to quickly copy their column title or **copy the underlying column prompt**. 


:::

:::VerticalSplitItem


![](https://api.archbee.com/api/optimize/A8WZzaoFAHNpve3k5PHtC/OyIXKHssw2yHdOO-UhLJr_image.png)
:::
::::

### Splitting Cells with Multiple Values

For certain column types, this menu will give users the option to **Split cells with multiple values. **This feature will explode each row into multiple, one for each distinct value in a given cell. One common application for this is to split unique product-indication combinations into individual rows as follows: 

| **Product** | **Indication(s)**  |
| ----------- | ------------------ |
| Product A   | Indication 1, 2, 3 |

**Becomes:**

| **Product** | **Indication(s)** |
| ----------- | ----------------- |
| Product A   | Indication 1      |
| Product A   | Indication 2      |
| Product A   | Indication 3      |

**Note:** This split functionality will occur automatically when certain columns are added. For example, adding a Dosage will automatically split the indications into unique rows to associate each with the corresponding dose. 

### Generate Filtered Cells

If you unfilter columns after having already generated new columns on your filtered table, you will notice the cells from these previously filtered columns to be empty:

::Image[]{src="https://api.archbee.com/api/optimize/A8WZzaoFAHNpve3k5PHtC/7cT-Jup3FM97OiTI0bpfT_.blob" size="56" width="393" height="631" position="center" showCaption="false"}

You then have the option to generate all missing cells by right clicking. This will generate a cell value based on the column prompt for these cells:

::Image[]{src="https://api.archbee.com/api/optimize/A8WZzaoFAHNpve3k5PHtC/bbPiRpON7vOuNG8tg5ihX_.blob" size="50" width="574" height="569" position="center" showCaption="false"}