What's New: MindsDB March 2020 Product Updates

Zoran Pandovski

Take a look at what we've accomplished in the month of March.

Welcome to your comprehensive look at what’s new in MindsDB for March 2020. 

MindsDB Scout

We are pleased to announce the beta launch of the newest version of MindsDB Scout. This latest version, MindsDB Scout 1.0.0, includes:

The updated Connection dashboard offers users easy navigation through the Local and Remote Server connections.

Connection Dashboard

The new Datasources dashboard provides Quality Score insights of the data where the user can get data columns scores for completeness, correctness and other statistics of the data.

Datasource Quality

The redesigned Predictors dashboard aims to answer the most important questions related to the generated models:

1. Is model complete

2. What is relevant for this model

3. When can you trust this model

Predictor Quality

By previewing the answers from the above questions, users can visualize the   model accuracies, column importances, confusion matrix. Now, Scout’s Predictors dashboard offers an advanced mode where users can add additional arguments as the margin of error, use GPU, choose backend, columns to be removed, etc before training the model.

Predictors Advanced Mode

In the new Query dashboard, the user can get predictions from the models. Scout’s new Query dashboard offers an explanation and extra insights about each prediction.

Query Prediction Dashboard

 We’ll release a new tutorial where you can learn all of the new MindsDB Scout features soon.

MindsDB

Recently  released MindsDB version 1.14.1 includes:

  • Improved confidence when Ludwig is used as a backend
  • Minor improvements to the model analysis results
  • Improved MindsDB CI build status to show status per supported Operating Systems (Linux, macOS, Windows)
  • More detailed confidence info by using TransactionOutputRow explain function that will return prediction quality, confidence interval and also important missing information.

Lightwood

The latest Lightwood version 0.18.1 includes:

  • A new numerical encoder that will improve the model accuracy
  • Range predictions

MindsDB Server

The MindsDB Server has a new endpoint for renaming the models. A simple example for invoking the new endpoint with python is:


The same new endpoint is available in MindsDB JS SDK too.

MindsDB JS SDK

We have released MindsDB SDK for JavaScript on npm. You can install it by using 

MindsDB Live Machine Learning Classes

As in February, in March we have continued hosting weekly Live Machine Learning classes. The aim is to help developers learn the fundamentals of Machine Learning in an hour long  interactive session that includes a  question and answers portion where attendees can get real-time answers from our team.

Our next class will be on March 26th, 2020 (Thursday) at 12:00 PM (PDT). You can register to join us for it here

Author Bio

Zoran is a full stack developer based in Macedonia. He works as MindsDB's senior full stack developer and works on everything from building and managing the website to supporting the open source product to working with users on their support questions.

Be Part of Our Community.

Join our growing community.