Flek opens the door on a world of innovation in model learning, exploration and prediction.
Unique Features

Flek offers a series of unique features specifically targeting modelling, exploration and prediction. Thanks to its self-adjusting universal probabilistic models users spend little time training or retraining their models as new unseen data comes in or the target variable changes when running predictions.

At its core is the powerful Probability Machine and FlekML Engine that allow users to:
-
Auto-learn semi-supervised models that self adjust with new data, while needing little training and tuning.
-
Model complex events that do not fit any known probability distribution.
-
Auto-discover interesting rules, associations, influencers, anomalies, polymalies or causal relationships.
-
Find hidden probabilistic patterns by searching the full joint and conditional probability distributions of various variables.
-
Develop ML-driven applications that dynamically query and mine the learned probabilistic models.
-
Run both forward and backward prediction and classification activities without retraining or remodelling.
-
Perform a combination of both profile and item based recommendation tasks
-
Run machine learning in a database-like pipeline that serves models similar to the way models are built and shared in SQL databases today.
-
Perform Algorithmic Probability Programming (APP) - a new paradigm that enables users to work with probability at a higher level of abstraction.
-
Easily interpret and visualization the results of probabilistic reasoning and trace the rules used during inference.