Please consider that ChatAible is a hosted service and avoid uploading potentially sensitive data or PII (Personally Identifiable Information).
If you upgrade to a paid Aible subscription we will configure a separate Aible environment linked to own cloud account. At that point when you load data it goes directly to your cloud account - bypassing Aible altogether - and all data storage and processing will occur in your cloud account. Reach out to Aible Customer Success to learn more about Aible’s security model.
Need some data to test with? Try out our public datasets to get started.
To upload your data to ChatAible click Upload CSV:
ChatAible uses Aible Sense to prepare and analyze your data.
When you upload a file Aible Sense first scans the data to verify that it is a valid CSV or gzipped CSV. Aible automatically fixes any minor problems - for instance missing values are not a problem but are explicitly converted to a dedicated “Missing” value.
The initial parsing analysis normally takes a couple of minutes. At this point your dataset card shows up in grey:
(Occasionally a file may have problems Aible can’t fix automatically in which case it highlights what needs adjusting before the data can be read).
There is just one more step to complete before you ask ChatAible about your data. You need to indicate which column to use as the modeling objective. This is a KPI you would typically plot on the vertical axis of a chart or a binary outcome a classification model would predict.
If you select a numeric field - indicated by the 123 to the right of the selection - Aible Sense will apply a regression analysis and identify patterns in the data that drive the value.
If you select a categorical field - indicated by abc - Aible will apply a binary classification to identify patterns in the data that drive the outcome. If your categorical field has more than two values you specify which is the target, positive value and all other values are grouped together as the negative value.
Your dataset can contain several different KPIs or outcome fields - you can change the target outcome and re-analyse the data at any point.
On the other hand, if you don’t have a suitable outcome column in your data you can create one by applying a transformation - typically using a Conditional Derived Field. We’ll cover transformations more in a separate topic.
Once you’ve selected the outcome column it takes a few more minutes to evaluate your dataset. Aible analyses your data to identify which columns, values and interactions of columns and values are good at explaining the variance in your outcome field.
We’ll explain the data readiness score and other analysis output in another topic.
Your dataset with its selected outcome column is represented as a green card in the datasets view. If you change the outcome Aible will create a new card for the new analysis. The same uploaded CSV can be represented in multiple cards each with different transformations and outcome analyses.
Once your card is green you are ready to analyze using ChatAible - click on the Ask ChatAible link at the top to get started.
Data is limited to 500,000 rows during the Beta release.