Beta This is a beta feature, which means we’re still developing it. Some functionality might change.
Data Analyzer is Quickbase’s built-in automated machine learning capability that delivers deep, actionable insights for your business, by finding statistical trends and correlations in your data. Data Analyzer lets business experts access the power of previously out-of-reach data science tools to drive better decisions based on the data already in their Quickbase applications.
This feature is available for Business and Enterprise plans only.
Learn about the Data Analyzer in five minutes
Data Analyzer consists of three components that work together:
- the Data Analyzer Template App that you get from Exchange
- the Data Analyzer plugin/wizard that coordinates the process.
- the Data Analyzer AI that generates models in the form of records in the app
As a feature, you'll use Data Analyzer in this flow:
- Understand the feature and how it can benefit you
- Identify your desired business outcome and how it translates into Quickbase data
- Locate, prepare and load data into the Data Analyzer Plugin/Wizard
- Read your created model and reiterate if necessary
- Take action!
More detail on Data Analyzer installation and use
Get the Data Template Analyzer app
Your first step will be to get the Data Analyzer Template app, explore the sample data and models.
How to read the Model Record
When reading the model record, you should remember what the Data Analyzer AI is doing. It is looking for statistical correlation between an outcome (i.e., a value in a field) and most common values in other fields in records that have the set outcome. The Data Analyzer AI does not understand what’s in the fields, apart from if it’s numbers, text or a checkbox.
Each model is a record in the ML Models table. Each successfully generated model will come with a form containing information and reports.
These are the main model attributes:
- Record ID of the Model in the table and the Alias you’ve given it upon creation
- Timestamp of when it was started
- Table name that contains the records that were analyzed
- The outcome that you’re predicting for
- Outcome rate (all records) — % of the records in the loaded report that have the outcome.
- # Fields used in prediction — the model is taking these fields in consideration. As the model is purely statistical, you can expect that not all fields play a role in predicting the outcome.
- ROC score — probability of the model predicting the outcome in two random records with opposite outcomes correctly. The higher, the better.
- Highest Probability — This chart is comparing the probability in your top 1% of records versus all of the records. This represents a small overview on how the model performs at its best and on average. The bigger the difference, the better the model is throughout all the analyzed records.
There are three tabs - Insights, Details and Deployment.
- Insights contains the list of insights and factors you have in your data based on the outcome. They follow a general format of “When “some field” is “with value X”, your outcome is “MORE/LESS” likely to happen.” In the end, we show the correlation, which can be a negative or positive number between -1 and 1. The closer the correlation number is to these values, the stronger the factor is when determining the outcome.
- Model Details contains the list of factors that are significant enough for the outcome. Some of the factors that you saw in the Insights are not here. It’s not because they are not true, but they are statistically redundant.
- Model Deployment contains the model converted to a Quickbase formula, which if you decide to deploy, will give a percent prediction for the outcome on each record in your origin data.
Install the Data Analyzer plugin
Now install the Data Analyzer plugin/wizard. See Data Analyzer Installation.
Using the Data Analyzer AI to generate models
With the Data Analyzer Template App and the Data Analyzer plugin/wizard installed, you can use the Data Analyzer AI to generate models in the form of records in the app. See Using the Data Analyzer AI to get started.
Tips to using Data Analyzer
For tips to using Data Analyzer, see Data Analyzer Tips.