A tool for performing exhaustive search of your data and entirely automating the creation of Linear and Logistic Regression Models:

Exhaustive Jube.io models are better than a manual approach because:

  • Jube.io Supports: Exhaustive Jube.io is very simple to use,  however our analysts are here to support you in its use and maintain an ongoing relationship.  Automation aside,  the support offered by our analysts will help you optimize the Exhaustive Jube.io settings and make sense of its output for the purposes of deployment.   It is our ongoing support that clients value the most.
  • Exhaustive Jube.io Cleans: Exhaustive Jube.io will clean the data removing outliers where appropriate and / or upon request.  Furthermore,  in the case of classification,  it will automatically assure that the dataset is symmetric around the dependent variable.
  • Exhaustive Jube.io Transforms: Exhaustive Jube.io will try thousands of variable transformations and compounded variables to obtain stark improvement on the correlations observed in the raw variables alone.
  • Exhaustive Jube.io Models: Exhaustive Jube.io will try thousands, if not millions, of Regression Models using different variable combinations to arrive at a genuinely optimal  model.
  • Exhaustive Jube.io Deploys: Exhaustive Jube arrives as an intuitive and explainable formula and suggests an appropriate activation threshold.

Features of Exhaustive Jube.io:

  • Symmetric Assurance:  Ensuring that for Classification models data is symmetric around the dependent variable, leaving Numeric Prediction models as is.
  • Raw and Best Transform Carry: Transforming the Independent Variables via simple statistical means and carrying forward for further analysis in the event that improvement has been obtained.
  • Outlier Removal: Removing Outliers above a specific number of Standard Deviations.
  • Abstractions:  Using all Independent Variables to perform an Exhaustive Search of Abstractions to improve Correlation of Raw and Transformed Independent Variables.
  • Correlation Disposal: Disposing of Independent Variables and Abstractions that do not meet a minimum correlation criterion based on a rolling number standard deviations.
  • Modelling Trials: Performing thousands, if not millions, of Linear or Logistic Regression models based upon the random selection of variables surviving Correlation Disposal.
  • Multicollinearity Disposal:  For each Modelling Trial, removing variables with high collinearity.
  • Max Entropy Activation Function:  For an optimal Logistic Regression model only, selecting an appropriate threshold to activate the formula output.
    While Exhaustive Jube.io is performing a lot of steps, invocation is just a few clicks.