Generality on Autoflow and AutoML 2.0
Present since 2018, the AutoML trend is an effort to make Data Science and Machine Learning open to the public.
The concept of automated/augmented AI models is in high demand, especially by enterprises. From now on, Artificial Intelligence is no longer just for experts. With AutoML, the true democratization of AI, all data science processes are accelerated and performed on our behalf. SmartPredict Autoflow is actually based on AutoML 2.0- the most recent version of AutoML-that offers an even higher level of automation by integrating pipeline creation. What really set SmartPredict Autoflow apart is the ability to be configured even after deployment.
Autoflow VS Traditional Data Science process
Autoflow processes all the data science tedious tasks in one flow to become a high-level automated process. Ranging from feature engineering to fine-tuning of hyperparameters, Autoflow includes all the data science processing pipeline stages, so all the user has to do is upload a well-formatted data, precise the target of prediction, then just let Autoflow fulfill the task of treating it.
- Auto Preprocessing
- Auto Feature selection
- Auto Feature engineering
- Auto Model selection and tuning of hyperparameters
- Auto pipeline generation
What distinguishes Autoflow from other solutions?
The majority of competitors tend to deploy models without a configurable pipeline. So we saw this as an opportunity to make DSML even more accessible. By offering a unique feature allowing pipeline configuration, this is exactly what we decided to offer our users: the ability to customize at will with Autoflow.
SmartPredict can be proud to be up to date on the state-of-the-art of Data Science. Based on the following cutting-edge open source projects, Autoflow offers an upgraded version combining all their advantages while mitigating or eliminating their potential weaknesses.
A modular approach coupled with automatic pipeline flowchart generation
The Autoflow approach is a derivation of the modular approach we have chosen to represent our ML pipeline. In short, we offer users the ability to automatically generate a flowchart of these modules that they can configure according to their preferences later on.