# How to test data in VFL?

**URL:** <https://discuss.flower.ai/t/how-to-test-data-in-vfl/192>\
**Category:** Flower Framework\
**Created:** [April 8, 2024, 8:12am UTC](https://discuss.flower.ai/t/how-to-test-data-in-vfl/192 "2024-04-08T08:12:41Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![emon21999](https://avatars.discourse-cdn.com/v4/letter/e/49beb7/32.png) [@emon21999](https://discuss.flower.ai/u/emon21999)\
**Post date:** [April 8, 2024, 8:12am UTC](https://discuss.flower.ai/t/how-to-test-data-in-vfl/192/1 "2024-04-08T08:12:42Z")

</div>

I was following the VFL tutorial in YouTube. I was wondering how to test the model on test.csv data of the Titanic dataset?

In the tutorial, we only see the simulation to train the client and server model, but there is no segment to test the model on new data.

How to achieve that?

---

<div class="post-metadata">

**Author:** ![danielnata](https://dub1.discourse-cdn.com/flex013/user_avatar/discuss.flower.ai/danielnata/32/22_2.png) [@danielnata](https://discuss.flower.ai/u/danielnata)\
**Post date:** [April 12, 2024, 2:29am UTC](https://discuss.flower.ai/t/how-to-test-data-in-vfl/192/2 "2024-04-12T02:29:10Z")

</div>

Hi @emon21999,

as the clients do not hold any labels to test their local models, evaluation can’t be done locally. You can try to test the model on test.csv data using centralized evaluation.

---

<div class="post-metadata">

**Author:** ![emon21999](https://avatars.discourse-cdn.com/v4/letter/e/49beb7/32.png) [@emon21999](https://discuss.flower.ai/u/emon21999)\
**Post date:** [April 12, 2024, 3:45am UTC](https://discuss.flower.ai/t/how-to-test-data-in-vfl/192/3 "2024-04-12T03:45:24Z")

</div>

But test.csv has all the features which were split to send to separate models and the server model can only handle the generated 12 features. Then how to approach this?