Farrago is a platform that abstracts away the cost and complexity of using predictive analytics.
This means that you can start building predictions for your business that improve profit and reduce risk. A question many of our technical users ask is: that’s interesting but what does the product actually do? In this blog post, we answer that question!
Without the need of data scientist support, data is prepared in order to obtain predictive models that can be tested and deployed to production in less than one hour. Often, it only takes a few minutes to deliver a machine learning model that is automatically trained with your data.
The phases involved in building a predictive model that leverages machine learning are:
- Data preprocessing: Format data appropriately and make sure it is in a readable tabular format.
- Data analysis: Understand what predictive task can be achieved with the data and how data should be transformed for the predictive model that is going to be used.
- Data transformation: Transform the data according to what has been established during the previous step.
- Separate training & validation data: The best way to see if a model will work is by testing the predictions against real results. In order to train and validate the model accurately, the data needs to be split into separate subsets. The criteria for splitting the data is not random and depends on the kind of prediction and on the nature of the data itself. Farrago automatically understands how the data should be partitioned in order to validate the results. The data needs to be split into two groups: training and validation.
- Train model: Use part of the transformed data for training and obtain a model.
- Validate model: Use part of the transformed data not used in the training phase in order to check how the model performs against known predictions.
- Deploy model: Deploy the model in your environment so that new data can be fed in for getting new predictions.
- Double-check model: Use new data in order to see whether the model is accurate even with data from a different sample.
While the data preprocessing part mainly assesses the formatting of the data, the data analysis and transformation phases are crucial in order to have data ready for Machine Learning. Generally, this is a task that should be performed by a data scientist or a data engineer, as the transformations that need to be applied depend on the model and on the kind of data contained in the dataset. Farrago AI takes care of this phase for you in minutes, where an expert would take up to 2 weeks.
Deploying the model can be challenging depending on the IT architecture within your organization. The Farrago AI team has the technical expertise to set the technology up so that it can be automatically deployed.
We’ve endeavored to answer the “How does Farrago AI do it?” question above.
If you still have questions, please don’t hesitate to email email@example.com. We love engaging with our community and welcome opportunities to refine our learning content.