Business Models

Business Models are listed in the Solution page. A Business Model in TAZI represents a specific approach to the Solution you defined earlier. With each Business Model, you may use a different configuration scheme, meaning that you can change the data source that you want to feed to your model, apply different feature engineering techniques, use a different ML algorithm or combination of algorithms and so on.

A new Business Model can be created for each different set of parameters that you want to try out and compare the model outputs.

When you create/open a Solution, you will be directed to a page where you can create a new Business Model or modify the ones you have already created. All of your Business Models related to a particular Solution will be listed here:

By default, Business Models will be grid listed. You can change it to a list view easily by clicking the list view button:

Create a New Business Model

You can create a new Business Model by clicking the +Business Model button. When you hover over the button you'll see there are two options to choose from. You can either create a Quick or a Customized version:

With Quick Version, feature engineering/selection/imputation and all ML model configurations are set automatically with default parameters so that; data scientists, business analysts and executives who don't know much about machine learning can create a model quickly.

Customized Version, on the other hand, can be used by people who are more experienced when it comes to data science and machine learning algorithms. With Customized Version, all parameters of Business Models can be set manually.

Quick Version

When you click the Quick button, TAZI will ask you to choose from some basic options and the rest will be auto configured with default options. However, you can change them later if you decide to modify the model.

1.Select Data Source

You can select an existing data source or create a new one. Let's choose an already created one and click Next. For more details on data source creation, click here

2.Define Target

Now, you will select your target feature that you want to predict. You can leave the other options as default as we will explain them later when we create a customized Business Model. Just see to it that you choose your problem type correctly depending on your use case. Here, classification is selected. Click Next to go to the final step.

3.Select Destination

In this step, you will choose an output destination to save your model outputs. You can create a new destination or select an existing one. Let's choose the default option and click Submit. For more details on Output Destination, click here.

When you click the Submit button, TAZI will automatically preprocess your data and create your Business Model.

Customized Version

You can create a customized Business Model by selecting the Customized button when you hover over + Business Model button or click the + Business Model right away. In the customized version, there are more options to choose from. But it is still possible to select default options and just click Next. Let's explain the steps:

Connect Data Source

You can either create a new data source on the go while creating the business model or select a data source that has been created before.

Let's choose a data source that's been already created:

We can edit the data source by clicking Edit Data Source button below. In the opened window, you can change the data source name, data source description and chosen delimeter for your file input. You can even drop a new data source file. Please note that this window would be different if we chose a database table for data source. See Data Source to understand how to use different types of data sources other than flat files. Now, let's close the Edit Data Source window and click Next.

Calculate Advanced Statistics

You can choose an already created data source or upload a new one for Characterizer which preprocesses the dataset by analyzing it statistically before feeding it to the machine learning model. By default the same data source is used to calculate the statistics. However, if your dataset is too large you can use a sample of your data to give it to the Characterizer and let it do the preprocessing by using a certain part of your dataset. This will shorten the running time of your model, indeed if you are dealing with a large dataset.

Let's choose the same data source since our dataset is not that large and click Next.