System and method of training machine learning models to generate intuitive probabilities

A system for training a machine learning model for predicting probabilities for binary outcomes to automatically generate at least one positive class prediction close to an ideal one probability and at least one negative class prediction close to an ideal zero probability, the system comprising:

• a memory that stores a set of instructions and an information associated with a machine learning algorithm;
• a processor that executes the set of instructions, wherein the processor is configured for training the machine learning model for predicting probabilities for binary outcomes to automatically generate the positive class prediction close to one and the negative class prediction close to zero, wherein the training comprises:
• generating a predictive probability curve based on the machine learning using a training dataset;
• and
• transforming the predictive probability curve into a transformed probability curve with probabilities spread close to the ideal one probability for a plurality of positive class predictions and close to the ideal zero probability for a plurality of negative class predictions, thereby introducing a valley in the transformed probability curve separating the plurality of positive class predictions from the plurality of negative class predictions, wherein a positive class prediction indicates a successful prediction case and a negative class prediction indicates a failed prediction case, and wherein the transforming comprises one of:
• 1) minimizing a distance between:
• a) the ideal one probability and a predicted probability value of the plurality of positive class predictions;
• and b) the ideal zero probability and a predicted probability value of the plurality of negative class predictions;
• and
• 2) maximizing a distance of a plurality of predicted probability values from a center of the predictive probability curve;
• wherein the trained machine learning model facilitates in real-time at least one of:
• 1) enabling at least one automated workflow, based on one or more rules conditioned on a distribution of the predicted probabilities obtained from the trained machine learning model;
• and
• 2) correctly classifying the plurality of predicted probabilities obtained from the trained machine learning model and presenting the plurality of correctly classified predicted probabilities on a display device.

A processor-implemented method for training a machine learning model for predicting probabilities for binary outcomes to automatically generate at least one positive class prediction close to an ideal one probability and at least one negative class prediction close to an ideal zero probability, the method comprising:

• generating a predictive probability curve based on a machine learning algorithm using a training dataset;
• and
• transforming the predictive probability curve into a transformed probability curve with probabilities spread close to the ideal one probability for a plurality of positive class predictions and close to the ideal zero probability for a plurality of negative class predictions, thereby introducing a valley in the transformed probability curve separating the plurality of positive class predictions from the plurality of negative class predictions, wherein a positive class prediction indicates a successful prediction case and a negative class prediction indicates a failed prediction case, and wherein the transforming comprises one of:
• 1) minimizing a distance between:
• a) the ideal one probability and a predicted probability value of the plurality of positive class predictions;
• and b) the ideal zero probability and a predicted probability value of the plurality of negative class predictions;
• and
• 2) maximizing a distance of a plurality of predicted probability values from a center of the predictive probability curve;
• wherein the trained machine learning model facilitates in real-time at least one of:
• 1) enabling at least one automated workflow, based on one or more rules conditioned on a distribution of the predicted probabilities obtained from the trained machine learning model;
• and
• 2) correctly classifying the plurality of predicted probabilities obtained from the trained machine learning model and presenting the plurality of correctly classified predicted probabilities on a display device.