# Support Vector Machines

**URL:** <https://datashield.discourse.group/t/support-vector-machines/1221>\
**Category:** Statistical development\
**Tags:** software-release\
**Created:** [14 March 2025 13:37 UTC](https://datashield.discourse.group/t/support-vector-machines/1221 "2025-03-14T13:37:50Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![gospelON](https://avatars.discourse-cdn.com/v4/letter/g/6de8d8/32.png) [@gospelON](https://datashield.discourse.group/u/gospelON)\
**Post date:** [14 March 2025 13:37 UTC](https://datashield.discourse.group/t/support-vector-machines/1221/1 "2025-03-14T13:37:50Z")

</div>

Please, Is there an implementation of a Support Vector Machine (SVM) package in Datashield? or is someone currently developing SVM for Datashield?

---

<div class="post-metadata">

**Author:** ![Jim-smith](https://avatars.discourse-cdn.com/v4/letter/j/ac8455/32.png) [@Jim-smith](https://datashield.discourse.group/u/Jim-smith)\
**Post date:** [14 March 2025 13:59 UTC](https://datashield.discourse.group/t/support-vector-machines/1221/2 "2025-03-14T13:59:17Z")

</div>

Given that support vector machines directly encode training data as part of the model this would be hard - unless you can embed the data in something like a differential private embedding first. There’s an example in the SafeSVC class in in the sacro-ml repo. But you would still have the issue of checking against class disclosure which ML models are as prone to as other forms of analysis

---

<div class="post-metadata">

**Author:** ![olly](https://yyz2.discourse-cdn.com/free1/user_avatar/datashield.discourse.group/olly/32/33_2.png) [@olly](https://datashield.discourse.group/u/olly)\
**Post date:** [15 March 2025 19:10 UTC](https://datashield.discourse.group/t/support-vector-machines/1221/3 "2025-03-15T19:10:08Z")

</div>

Moved thread to stats dev category.

---

<div class="post-metadata">

**Author:** ![gospelON](https://avatars.discourse-cdn.com/v4/letter/g/6de8d8/32.png) [@gospelON](https://datashield.discourse.group/u/gospelON)\
**Post date:** [17 March 2025 18:11 UTC](https://datashield.discourse.group/t/support-vector-machines/1221/4 "2025-03-17T18:11:36Z")

</div>

Ok, thank you Jim-smith for suggesting SafeSVC.

---

<div class="post-metadata">

**Author:** ![florian.schwarz](https://avatars.discourse-cdn.com/v4/letter/f/73ab20/32.png) [@florian.schwarz](https://datashield.discourse.group/u/florian.schwarz)\
**Post date:** [17 March 2025 18:21 UTC](https://datashield.discourse.group/t/support-vector-machines/1221/5 "2025-03-17T18:21:52Z")

</div>

Vielen Dank für Ihre Nachricht! Ich bin vom 17.03.2025 bis 21.03.2025 im Urlaub und daher in dieser Zeit nicht erreichbar.

Mit besten Grüßen Florian Schwarz

Thanks for your message! I am currently on holidays from 17/03/25 until 21/03/25 and therefore not available.

Best regards, Florian Schwarz

> > > Gospel Ozioma nnadi via DataSHIELD [notifications@datashield1.discoursemail.com](mailto:notifications@datashield1.discoursemail.com) 17.3.25 19:21 \>\>\>

> ![](https://global.discourse-cdn.com/free1/uploads/datashield1/original/1X/35da766bf7fc49e5517467fe753da12fe6d61814.png "gospelON") | [gospelON](https://datashield.discourse.group/u/gospelon)  
> 17 March |
> 
> - | - |

Ok, thank you Jim-smith for suggesting SafeSVC.