TADA^® Augmented Analytics for Every Professional

By clicking on the “View Data” button, you can access the graphical visualisation of your data. There are different graphs depending on the type of variable (numeric, categoric, text).

A goal is the variable that you want the model to predict. Click on “Next: Create Goal” to get started creating a goal. 

First, you need to select the “Goal” variable among the variables from your dataset. Once selected, you can visualise again the data repartition graph for the selected variable.

Click on “Next: Create a variable set” to get started creating your variable set. A variable set includes the variables that are taken into account during the model generation process. The variable chosen as the Goal cannot be included in a Variable set. 

Here, you need to select variables to be explored. If needed, you can launch a Variable Reduction to create a smaller variable set with the most relevant variables.

To Generate predictive models, you can simply click on the «Generate a model» button. 

You get to this new screen where you see a box for each individual model generated. At the beginning there is only one box. If you want to generate many models, click on “generate a model” as many times as you want an additional model to be generated. 

Then, you can visualise all the predictive models of a variable set and quickly compare them with the Global Score, an indicator of the performance of your models.

Click on the box representing the model you want to understand. 

First, the rainbow type display shows the model’s Global Score, which is a summary of the model’s performance. Then, the advanced metrics provide three performance indicators known in data science. Finally, the global influence diagram shows a sensitivity analysis.

The Actual vs Predicted is a visual representation of the results of your model (the expected values in pink and the predicted values in blue). The closer the dots are to the line, the better the model. In the Metrics section you’ll find more data science metrics to assess the quality of your predictive model.

Local influence is an analysis of the impact of the values taken by each variable of the model on the predictions, in order to better trust the model and the machine learning process, identify trends in data, and improve the understanding of the business reality.

With the Live Predict, you can quickly and easily adjust the values of variables, single or combined, and see right away the corresponding impact on the goal prediction.

You can explore scenarios and make decisions or recommendations that apply to business reality.

First, browse or drag and drop the dataset on which you want to apply the model. Then, the model will be applied to the entire file and a new column will be added to your file with the predictions. Finally, you can download the new file containing the new column with the prediction.

Once generated, the predictive models provided by TADA are yours! You thus obtain the model as a mathematical formula. In addition, TADA automatically implements the code of your model. Choose C++, Python or Javascript and get your code ready to be integrated into your own applications.