Creating models (data source) and experiments

<< Click to Display Table of Contents >>

Navigation:  Cloud applications > Bizagi Artificial Intelligence >

Creating models (data source) and experiments

Overview

Artificial Intelligence for Bizagi Cloud, is an application which allows you to explore machine learning capabilities directly in your Bizagi processes, by having processes rely on a predictive analysis service that is based upon the reliable data that you have in a Bizagi Dataset.

Throughout predictive analysis, you will be able to train models and carry out experiments that rely on renown machine learning algorithms to determine a given outcome based on stored data (with a given certainty).

For such models, you may easily connect your Bizagi processes so that these processes present a prediction once certain data has been inputted (to either set as default value or to suggest to the end user).

For introductory information on this application, refer to Bizagi Artificial Intelligence.

 

This section describes how to get started with Artificial Intelligence, and create AI models and experiments.

 

AI models and experiments

In order to work with Artificial intelligence and its capabilities for Bizagi Cloud, you need to first define the data source from which AI services will work on.

Note that as it is usual in machine learning fields, a model will be built based on sample inputs.

 

Defining that data source is done by creating a model in Bizagi Artificial Intelligence's UI.

For that model, you may then create any number of associated experiments, so that within each experiment, you may explore different results (evaluate possible combinations between input parameters and algorithms).

 

The results you get, will be in terms of a suggested algorithm for use case statement, and giving you a certainty for that given output (attribute) you want to predict.

For the prediction you define as input parameters. those attributes you identity as relevant.

 

The following image illustrates this concept:

 

Cloud_AI_experiment

 

This means you may continue to create further experiments for different trials aimed at improving the certainty of the prediction so that it is better in new experiments.

Note that you need to know and specify beforehand, which attribute you are seeking to predict (a supervised learning treatment).

 

Prediction certainty

An AI experiment will present a prediction's certainty or precision.

Such precision is given either as an accuracy percentage or as a standard error value, according to the data type of the attribute you want to predict:

 

Accuracy

The prediction's accuracy can be seen as the number of correct predictions divided by the total predictions made (then multiplied by 100 to be turned into a percentage).

Determining if a given accuracy is appropriate or not, depends on each specific use case.

However it is usually common that an accuracy higher than 70% can be considered as good/reliable (the higher the accuracy, the better).

 

Such accuracy applies to attributes whose data type are categorical.

Examples of categorical data are:

oTypes of a rock:  igneous, sedimentary or metamorphic.

oA named geographical place where people live (country, state or city).

oA medical condition which is diagnosed as true or false.

 

Standard error

The prediction's standard error can be seen as the measure with which a value represents the complete population or data.

Determining if a given standard error is appropriate or not, depends on each specific use case.

For such decision, it is strongly recommended to analyze the data distribution of the attribute to predict.

The lower the standard error, the better.

 

Such standard error applies to attributes whose data type are continuous.

Examples of continuous data are usually numeric values, such as:

oA duration (e.g a measured elapsed time in seconds).

oA countable quantity of items.

oThe price of an item.

 

note_pin

Even though you should aim for a good prediction certainty, this still does not mean by itself that your model is exempt of having high variance or bias.

Recall that you should ensure that your data is final, reliable and that it does not need further preparation.

It is usually recommended to have large amount of information in the Dataset in order to produce reliable experiments.

 

Procedure

In order to create a new model, follow these steps:

 

1. Log in to the Management portal.

In order to log in, recall that you will need an account at www.bizagi.com which belongs to your corporate subscription, as created by your Cloud subscription owner, and as described at User access and registration.

 

Cloud_AccessCommon

 

2. Enter into your subscription details.

Once you log in, at the landing page you will see a list of your current subscriptions for you to click on the desired one.

 

Cloud_ManageCommon

 

3. Go into artificial intelligence page.

At the list of projects in your subscription, go to the Artificial Intelligence tab and click on Go to Project Board.

 

Cloud_AI1

 

4. Create a new Experiment.

Click on New Experiment in order to create an AI Experiment:

 

Cloud_AI4_new

 

The Experiment Wizard is opened, click on Let's Go!.

 

Cloud_AI5

 

4.1 Create an A.I. Project.

Create an AI project to help you classify and organize your further AI elements.

Ensure you give the AI project a Name and a meaningful Description, and click Next > when done.

 

Cloud_AI._projectInfo_new

 

Note that a AI project is not related to a Bizagi project holding processes.

Such AI project's main purpose is to help you organize and sort out the set of AI models and experiments you may have.

 

4.2 Create an A.I. Model.

Ensure you give it a Name, a meaningful Description, and configure the following adequately:

Datasource: Select the dataset containing the data you want to use contained in a dataset project.

Environment: Select the specific environment of the dataset having the data you want to use.

Notice you should always select a dataset with data from a production environment.

Otherwise experimenting with non-real data may yield results which aren't as applicable as in a real use of applications.

 

Cloud_AI_createModel_new

 

Click Next > when done.

 

Recall that you will be setting the data source from which sample data is taken.

Note that you are presented with those Dataset projects, Datasets and their environments, which belong to your same Bizagi Cloud subscription.

 

The image below shows the Dataset project and Dataset as employed for the AI model's data source:

 

Cloud_AI_datasets

 

6. Create A.I. Experiment.

Finally, create the experiment. Give it a Name and a meaningful Description.

 

Cloud_AI_createExp2_new

 

Click Finish > when done.

 

Cloud_AI6

 

At this point, your experiment is ready to start predictions.

To learn how to configure the experiment and work with its results, refer to Working with AI experiments.