Data Analyzer

Data Analyzer is an AI tool that predicts the likelihood an outcome will occur for each record in your table. Generate predictions to analyze Quickbase data, and view results directly in your table to better understand trends and drive decisions.

Note

Quickbase previously offered an Exchange app called Data Analyzer. With the release of the built-in Data Analyzer experience, this app is no longer offered. For questions on the app, reach out to technical support.

How it works

Data Analyzer reviews all the records in your Quickbase app to find patterns that help predict your outcome. It identifies specific details or conditions within your data that have a strong relationship to that outcome—for example, whether the Notes field contains the word “downgrade” or whether the Assigned to field isn’t empty. Using these observations, Data Analyzer builds a predictive model tailored to your app’s data.

Once the model is created, Data Analyzer runs it against all target records in your table to generate a prediction for each one. These predicted values are automatically added to your table in a new field, where you can use them in workflows, data visualizations, and other analytics.

Requirements

Permissions

To use Data Analyzer, you must have:

• App admin permissions in the app you want to generate predictions for.

• The Artificial Intelligence (AI) setting enabled. A realm admin can enable this from the Policies page in the Admin Console.

Tables

The table you generate predictions for must have:

• A minimum of 20 records

• A maximum of 330 million cells (records x fields) per table

• At least 5 fields (excluding the field you want to evaluate)

• At least 8 records where the outcome is true and 8 records where the outcome is false

Limits

Data Analyzer isn’t available for:

• Apps with sandbox enabled

• When testing as a role or a user

Predictions can only be made for these field types:

• Checkbox

• Text

• Multiple choice

Access Data Analyzer

Access Data Analyzer from app settings, or from the quick insights feature.

From app settings:

1. Select App settings from the sidebar

2. Under Advanced Features, select Data Analyzer

From quick insights:

1. Generate insights by following the steps in the quick insights article

2. Once your insights have been generated, select Open Data Analyzer
   This takes you directly to the modal to configure predictions, and prefills the inputs with the same table, field, and value to predict that you selected to generate insights.

Create a new prediction

Predictions show the percentage likelihood your selected value will occur for each record in your table. To create a new prediction:

1. In Data Analyzer, select New prediction. A modal opens.
   If you accessed Data Analyzer from quick insights, you’re taken directly to the new prediction modal.

2. Select the outcome you want to predict by choosing a table, field, and value to predict.

   The value must be:

   • An exact value, meaning the field type is either checkbox or text

   • Case-sensitive

   • Have at least 8 records where the outcome is true and 8 where the outcome is false. This means at least 8 records that contain the value, and 8 that don’t.

3. Enter a prediction field label. This label appears on a new field that’s added to your table to display prediction results.

4. Optionally modify default prediction behavior.

5. Select Generate.

Your results are emailed to you when they’re ready. This may take a few minutes.

Default prediction behavior

The default prediction behavior settings are used by Data Analyzer to achieve the most accurate predictions. By default, the values used are determined automatically based on your data.

Warning

Modifying these values may impact the accuracy of predictions

• Record creation date—The date field that best represents when the record was first created.

  For example, if you imported a set of records into Quickbase from a third-party system, the default Date Created field reflects the import date. Instead, select the date the record occurred, like date of sale, or date of project.

• Time it takes for the outcome to occur—The number of days it takes for almost all records to contain the selected value, or remain without it.

  For example, if you’re predicting the likelihood a sale will be confirmed, enter the number of days that nearly all records will take at your company for a sale to go through.

• Don’t generate predictions for records newer than—The number of days Data Analyzer excludes new records from the dataset for. Records that are too new often lack enough data to make reliable predictions.

View your results

Overview of results

From Data Analyzer, select your prediction from the table to view an overview of the results. You’ll find:

• Distribution of records by prediction likelihood: A graph showing the number of records compared the the percent likelihood they are to contain the value you selected.

• Accuracy: The expected accuracy based on the records analyzed in your table.

• Historical outcome rate: The percentage of records that contained the value you selected during a specific timeframe. Data Analyzer uses this timeframe to determine a baseline to compare to.

• Fields that impact predictions: A list of other fields across your app that had a significant impact on the prediction results.

Predictions per record

View the percentage likelihood your selected outcome will occur for each record in the new prediction field added to your table. This field is automatically added when predictions finish generating.

Update predictions

As new data is added, you may want to update predictions to ensure they are as accurate as possible, given the current data in your app. Updating clears the current results, and replaces them with new ones. It may take a few minutes for this process to complete, and you’ll be emailed when they’re ready.

Note: You cannot make changes to the configuration when updating predictions. To select a new outcome to predict, or modify the prediction behavior, create a new prediction.

To update predictions:

1. Access Data Analyzer from app settings

2. In the table view, select Update predictions

You can also select the prediction you want to update, and select Update predictions from the results page.