Revision Date:  27 March 2023

This documentation has been created for software version 2023.1.0.0

It is also valid for subsequent software versions as long as no new document version is shipped with the product.

Suite 105, 220 Davidson Ave, Somerset, NJ, USA 08873

Support: 

For more information, visit

https://www.whiz.ai/contact

Copyright © 2023 whiz.ai All Rights Reserved.

Trademarks owned by whiz.ai

Disclaimer

No Warranties and Limitation of Liability

Every effort has been made to ensure that this document is an accurate representation of the steps to deploy WHIZ.AI platform. However, the development of the software is a continuous process. So, small inconsistencies may occur. 

We would appreciate any feedback on this document 

Send comments via email 

To: support@whiz.ai 

Supported Browsers

WhizAI supports the following browsers:

• Google Chrome, v110 & above
• Mozilla Firefox, v110 & above
• Microsoft Edge, v110 & above (Windows); v106 & above (MacOS)
• Safari, v15 & above (MacOS)

What’s new

This document outlines the new product updates for the release version 2023.1.0.0. The following lists the new features and enhancements.

Pinboards, Cards, Responses, and Insights

• Response - Sankey Chart
• Response - Multi column horizontal bar charts
• Response - Bump Charts
• Response - Introducing Combo charts supporting trend lines
• Response - Arrow icon indicates the default sorting order
• Pinboards - Introducing Design Mode for Pinboards
• Pinboards (Cards) - Ability to save the sorting order on cards in pinboards
• Pinboards (Cards) - Ability to switch from Period over Period (POP) comparison to Year over Year (YOY) comparison on cards in pinboards
• Viewing Example Queries for each business category on the Info page
• Insights - Configuring factors for Key Drivers analysis
• Insights - Selecting the data model from the Templates tab for Anomalies and Key Drivers
• Insights - Introducing new options in the Anomaly dropdown list while creating narrative template for anomalies
• Insights - Query Suggestion
• Insights - Updated Key Drivers Analysis output
• Insights - Ability to support ‘Selected Period’ analysis for shorter period for key drivers analysis
• Insights - Performing the Period Over Period (POP) and Year Over Year (YOY) analysis for Key Drivers from the response
• NLQ Support - Word Sense Disambiguation
• NLQ Support - Specifying descending or ascending order in your queries
• NLQ Support - Supporting multiple dimensions for time comparison queries
• NLQ Support - Supporting time comparison for more than 2 time periods
• NLP - Slot filling enhancement
• NLP - Improved Natural Language Understanding (NLU)

Administration

• Data Modeler - Adding Entity Synonyms by uploading excel file from the Pipeline Manager
• Data Modeler - Adding Business Categories from the Pipeline Manager
• Data Modeler - Adding Example Queries to data model Info page
• Data Modeler - Configuring Calculated (Calc) metric from the UI
• Data Modeler - Using Data Dictionary UI to select the NLP Datasource for a dimension
• Performance Monitor - Generate audit logs on user creation, password change, and more
• Users & Security - Enhancements in user onboarding
• User interface (UI) improvements
• NLP Workbench - Description box for Replacement
• NLP Workbench - Adding synonyms for metadata
• Narratives - Viewing card level narratives on pinboards
• Narratives - Exporting narratives as a part of card
• Narratives - Building custom narratives just got better
• Narratives - Introducing building blocks to configure condition based narratives
• Narratives - Creating narratives around ‘Count’

Pinboards, Cards, Responses, and Insights

Response - Sankey Chart

 Now, you can view Sankey Chart visualization to view the flow of information.  Sankey charts represent the hierarchies and their levels.

For example, when you ask ‘Patient Count by lot1s, lot2s, lot3s,and lot4s’ then you get the response as shown in the following figure:

Response - Multi column horizontal bar charts

 Now for bar charts, data is displayed in horizontal bars with different colors (Please refer to the image). For this chart, categories of data are displayed on the vertical axis and the data values are displayed on the horizontal axis.

This way, this horizontal bar graph can be used to visualize comparison between various metrics and thereby analyze the data performance.

Example query: TRx, NRx, NBRx by regions

Limitations:

Response - Bump Charts

Now WhizAI offers another visualization option - Bump charts.  A Bump Chart is a simpler form of a line chart that can be used to view changes in ranks over time. With this chart you can easily compare the position or performance of multiple observations. It can be used to view how a particular region has performed on a defined sales metric over a period of time.

Example query: NBRx by Regions by months

Limitations:

• Trendlines are not supported when there are more than one metric and dimension.
• Roll Up/Drill Down is not supported. 

Introducing Combo charts supporting trend lines

WhizAI now supports combo charts, that basically are bar charts with a trend line. Combo charts enable you to analyze and compare historical data for two metrics. In the Combo charts visualization, you can see a dual axis bar chart, where each bar in the chart shows data of two metrics stacked on top of each other. Also, this chart shows a trend line  for the third metric for the time period included in the question.

Since values of metrics are combined in this chart, you can use these charts to view data comparisons.

Important! To view the combo-chart with a double axis bar, you have to include the necessary dimensions in your questions. Also, the questions must have one metric and multiple dimensions. 

Example query: 

• Trx and Nrx by month
• Months by Trx market share

To invoke the combo chart go to the info tab <configurable_name>_stack_chart 

Limitations:

• The filter for metric is not supported where there is a trend line option. 
• Roll Up/Drill Down is not supported as it is a multi-dimensional chart response.
• The chart does not support the option of xls/csv export.
• The granularity must be present in each question, and the granularity of each question must be the same.
• More than one question must be present in the NLQ field.

Arrow icon indicates the default sorting order

Now for table responses, either on explorer or pinboards, by default, you can see a small arrow icon ( ) besides the column name. This arrow indicates that when the response is generated, the data in that particular column is sorted, by default.

Also, this icon indicates the sorting order, that is, if the data is sorted in ascending or descending order. This allows you to intuitively analyze the data findings.

In the above image, by default, the data in Abs Chng column is sorted in descending order. To change it into ascending order, click the arrow.

Following are the default sorting conditions for different types of responses:

For example: Top districts by TRx

For example: TRx by Customer by product

Metric – TRx, Dimension - Customer, product

For example: Trx by region by product

Metric – TRx, Dimension - product, region

For example: Product by region by district

Metric – TRx, Dimension  - Product, Regions, Districts

For example: Arobi vs ofasan by district

Metric – Trx, Dimension - Arobi, Ofasan, Districts

Limitation:

Pinboards - Introducing Design Mode for Pinboards

Now, you can format your pinboards using the new design mode option. You can change the layout for:

• Pinboards and cards pinned to these boards
• Filters 
• Narratives on Pinboards

Changing the layout for pinboards:

From the Pinboard Layout tab, you can change the:

• Background color of the pinboard
• Font Color
• Font Size
• Font style of the pinboard name
• Rename the title of the boards.

 After you finish changing the layout, you can click the Save button to save all the changes.

Changing the Filters Layout:

Pinboard Level Narratives Layout:

On pinboards, when you click   hamburger menu > Design Mode option > Pinboard Level Narratives Layout tab. From the Narratives Layout tab you can Hide or Show all the narratives in the panel. Also, when you click Edit Narrative option you are redirected to the Narrative Template page.

From this tab, you can change: 

Ability to save the sorting order on cards in pinboards

Now, on cards in pinboards, when you sort the columns you can save that sorting order. For example, when you ask WhizAI ‘Show me TRx by region’ you get a response as shown in the following figure.

When you sort the column TRx and save the changes, the sorted order gets saved as shown in the following figure.

Ability to switch from Period over Period (POP) comparison to Year over Year (YOY) comparison on cards in pinboards

Now, on period over period comparison (PoP) cards in pinboards, you can switch to year over year (YoY) comparison. For example, when you ask WhizAI ‘Show me nrx, nbrx average by regions mtd pop’ you get the response as shown in the following figure.

Now, pin the response and expand the card and then click the hamburger button.

You can switch from POP to YOY growth and vice versa using the Period Operator toggle button as shown in the following figure.

Viewing Example Queries for each business category on the Info page

Now Example Queries on Info page are categorized according to the business categories on the Info page. For more information, refer to the following figures.

Example Queries for All business categories:

Example Queries for the business category Activity:

Example Queries for the business category Sales:

Insights - Configuring factors for Key Drivers analysis

WhizAI allows you to tailor a list of factors that should be focused upon while detecting the key drivers and contributors towards the performance of a given business metric. You can configure these factors from the Admin console > Insights Workbench. 

When you select a metric from Insights Workbench > Key Drivers tab, this list of factors is pre-populated for that metric.

To configure the list of factors:

The data displayed in the next two columns (Filters and Factors) is based on your selections in this column, this means, depending upon the Data Model you select, the filters and factors are accordingly displayed.

You receive a pop up message that the template has been saved successfully.

Now, when you select the metric NBRx from the Key Drivers tab the list of configured factors are populated as shown in the following figure.

Editing a template

To edit a template:

1. Click the Edit icon from the templates.
2. Edit the required parameters.
3. Click Save.

Deleting a template

To delete a template:

.

The following message pops up.

Insights - Selecting the data model from the Templates tab for Anomalies and Key Drivers

Now you can select the data model from: 

This way, whenever you want to view the templates for the different data models, you do not need to navigate to the General tab and select the data model. 

Limitations:

Insights -  Introducing new options in the Anomaly dropdown list while creating narrative template for anomalies

Now you can select the following options while creating a narrative template for anomalies.

• Deviation
• Smart Deviation
• Percent Deviation
• Smart Percent Deviation

Insights - Query Suggestion

Now WhizAI provides suggestions that complete a user’s search query. The Suggested Queries generates suggested search based on what you are searching for and the result within your dataset.

For example, when you ask WhizAI ‘Show me TRx by region’ you get the response along with the suggested queries as shown in the following figure.

Insights - Updated Key Drivers Analysis output

Now on the Key Drivers tab you have a new dropdown option Context to Analyze where you can set the context to run the analysis.  When you select the Context to Analyze and run the Key Drivers analysis, you can view the different output according to the selection of context. 

The Context to Analyze dropdown includes:

• Selected Period 
• Period over period change
• Year over year change

Key Drivers analysis output for Selected Period:

When you run the key drivers analysis for the context Selected Period, the performance of your metics is displayed for the entire selected period.

To view the top driving factors of your metric, for the context Selected Period, follow the steps listed below:

1.  Go to the Insights Workbench > Key Drivers tab and fill the details:

1. Metric : From this column, add the Data Model ( For example, Field Analytics ) Metric ( For example, NRx )

The Key Drivers analysis output displays the top driving factors of metric (NRx) for the selected period (last 52 weeks) as shown in the following figure.

The result is displayed into two distinct tabs:

1. Contributors: shows the significant contributors. 
2. Drivers: shows the key drivers.

Non Restricted in PDRP Flag with 85.2% and Practitioner in Customer Type with 74.3% contribution are discovered as some of the top contributors.

When you scroll down, you can view more information about each of the dimensions whose members are identified as significant contributors. 

The left-side view of the Contributors displays the absolute volume as shown in the following figure.

The right-side view of the Contributors displays the average performance for each of the dimension members as shown in the following figure.

The best fit line indicates the relationship of different data points on a scatter plot chart as shown in the following figure.

Key Drivers analysis output for Period over period change.

When you run the key drivers analysis for context period over period change, it compares the performance of the selected period with respect to the previous period and finds out the significant contributors towards the change in the performance.

To view the top driving  significant factors of your metric for Period over period change context, follow the steps listed below:

1. Metric : From this column add Data Model ( For example, Field Analytics ) Metric ( For example, NRx )
2. Select Context to Analyze ( Period over period ). 
3. Filters : In this column add filters to be applied on scope (For example, Period - last 1 months). 
4. Factors : In this column add factors to be used for potential key driver analysis. You can select the dimensions only.

In the Key Drivers analysis output, the Significant Contributors helps you find out the members whose performance has contributed significantly to the growth or decline in the metric.

WhizAI analyzes the performance difference between selected periods across the data, and displays significant contributors.

For example, tier 1 in Customer Tier with 38.87% and Cardiovascular Diseases in Primary Specialty Name with 24.33% contribution to the change are discovered as some of the top contributors. 

When you scroll down, you can view more information about each of the dimensions whose members are identified as significant contributors.

The left-side view displays the period over period difference for all the members of the particular dimension.

The right-side view displays the comparison between change in contribution between previous period and the current period.

Key Drivers analysis output for Year over year:

When you run the key drivers analysis for context year over year, it compares the performance of the selected period with respect to the previous year period and finds out the cause of the change in the performance.

To view the top driving factors of your metric for Year over year context, follow the steps listed below:

1. Metric : From this column add Data Model ( For example, Field Analytics ) Metric ( For example, NRx ) and Context to Analyze ( Year over year ). 
2. Filters : In this column add filters to be applied on scope (For example, Period - last 1 months ). 
3. Factors : In this column add factors to be used for potential key driver analysis. You can select the dimensions only.

In the Key Drivers analysis output, the Significant Contributors helps you find out the members whose performance has contributed significantly to the growth or decline in the metric.

WhizAI analyzes the performance difference between selected periods across the data, and displays significant contributors.

For example, year-over-year NRx grew by 2.99% in the period April 2022 as compared to April 2021.

The Cardiovascular Disease of dimension Primary Speciality is the top performing factor for the year over year change of the NRx as shown in the following figure.

The left-side view displays the year over year difference for all the members of the particular dimension.

The right-side view displays the comparison between change in contribution between previous year period and the current period.

Insights - Ability to support ‘Selected Period’ analysis for shorter period for key drivers analysis

Now on the key drivers analysis tab you can run the key drivers analysis for periods less than the threshold limit (36 weeks) for Context to Analyze Selected Period. 

Insights - Performing the Period Over Period (POP) and Year Over Year (YOY) analysis for Key Drivers from the response

Now, you can perform the period over period and year over analysis from the response table for the key drivers. Also, you can perform the Key Drivers analysis from the response table.

Period Over Period (POP) and Year Over Year (YOY) analysis: 

For example, when you ask ‘Show me NBRx monthly trend’, then you receive a response as shown in the following figure.

Key Drivers analysis:

For example, when you ask ‘Show me NBRx monthly trend’, then you receive a response as shown in the following figure.

Limitations:

• The POP or YOY analysis for key drivers on response is supported on single time series trend only.
• When no active templates are available for a given metric then you get the following message on mouse hover  “Analysis not supported as Key Driver template is not set for this metric. 
• For the “key driver analysis on selected period”, period over period (pop) and year over year (yoy) is supported only on the base metrics.
• Header name is not displayed in a separately opened tab while checking the insights for PoP analysis for Key Drivers from the response.
• The exported files contain only the visualizations and not the narratives displayed on the screen.

NLQ Support - Word Sense Disambiguation

Now, WhizAI can determine which meaning of a word is activated by the use of that word in a particular context. 

For example, consider two examples of distinct senses that exist for the word “May”.

Example 1: Show me TRx for May

Example 2: How much did May sell last year?

The word May denotes the distinct meaning. In the first sentence it means month and in second, it means the name of the person. Therefore, WhizAI disambiguates such queries using Word Sense Disambiguation (WSD) and gives the correct response.

Example 1: Show me TRx for May

Example 2: How much did May sell last year?

NLQ Support - Specifying descending or ascending order in your queries

 Now when you ask questions to WhizAI, you  can use the following phrases:

•  Descending order
•  Ascending order

In the response, WhizAI shows a list in the order that is specified in the query. 

Example queries:

• Show me brands in descending order by TRx
• Show me regions in ascending order by NRx

NLQ Support - Supporting multiple dimensions for time comparison queries

 Now, you can include multiple dimensions in time comparison queries. For more information, refer to the following example queries with responses.

Example query 1: PoP region by brand

Example query 2: YoY for regions by districts by territories

Limitation:

NLQ Support - Supporting time comparison for more than two time periods

Now WhizAI supports time comparison for more than two time periods. Meaning, you can compare data across more than two time periods. For more information, refer to the following example.

Example query:  Show me 2022 vs 2021 vs 2020 by months

Note! You can see the metric TRx is being compared across three different time periods

Limitation:

NLP -  Slot filling enhancement

We have improved our slot filling capability. Now, WhizAI suggests only those options that are relevant to the entities included in the query. For more information, refer to the following example. 

Example query: "Market share for AMARD"

Before:

All the available products are displayed as the slot filling options:

Now:

Since you asked for the market share for AMARD; instead of showing all the products, the products only for the AMARD market are displayed in the Product drop-down.

NLP - Improved Natural Language Understanding (NLU)

1.  We have expanded our natural language understanding coverage. Now, WhizAI understands your questions even when you ask the same question differently. Meaning you can frame one question using different variations of the sentence structure.  For example, if you want to know TRx sales in Q4 2022; you can frame your question in following different ways:

• Trx for Q4 2022
• TRx Q4 last year
• Last year Q4 for TRx

As shown in the table below:

Note! The purpose of this table is to provide, only, a few examples of how different variations are supported. This is not a comprehensive list

For example:

• Show me Plabenil vs Arobi this year - here ‘vs’ is used
• Show me comparison of plabenil and Arobi for this year - here we have included the word ‘Comparison’ since we are using ‘And’ in the question.
• Compare TRx for Arobi, Emarun
• Compare NRx trend for Ofasan, Trexine

Administration

Data Modeler - Adding Entity Synonyms by uploading Microsoft Excel file from the Pipeline Manager

Now, from the Admin console > Data Modeler > Pipeline Manager > Data Dictionary user interface,  you can add synonyms for dimension entities by uploading a Microsoft Excel (XLSX) file having synonyms for the dimension entities.

Note! Sheet name (Region) in the XLSX file is case sensitive and must match with the code of the dimension.

To add entity synonyms:

Excel file gets uploaded.

5.  Click Save Changes.

6.  Go to the Pipeline Manager page and run the pipeline.

After successful pipeline run, synonyms for the entities from the MS excel file are added to WhizAI.

Data Modeler - Adding business categories from the Pipeline Manager

Now, from the Admin console > Data Modeler > Pipeline Manager > Data Dictionary user interface,  you can add and configure business categories to include specific metrics and dimensions according to different business areas. These business categories are displayed on the data model Info page.

To add and configure business categories:
 

Following Business Category dialog opens.

After a successful pipeline run, the business category that you added (e.g: Sales) to the data model Info page.

Data Modeler - Adding Example Queries to data model Info page

Introducing Example Queries in the Data Modeler. You can use this option to add example queries on the data model Info page. You can add a single query at a time or you can add multiple queries by importing an excel file having multiple queries.

To add a single example query:
 

Query gets added for the data model.

Query gets added to the data model Info page as well.

To add multiple example queries:
 

1.  Go to the Admin console > Data Modeler > Example Queries.

2.  Click the menu icon and then Import.

Close the Import Example Queries dialog.

Queries get added to the data model Info page as well.

Limitations:

For this release:

• Example Queries are supported in English only.
• You cannot export the Example Queries.

Data Modeler - Configuring Calculated (Calc) Metric from the UI

You can now configure calculated metrics (referred as Calc Metric) from the Admin console > Data Modeler > Configuration page.  Calculated Metrics (or Calc Metrics) are user-defined metrics that are computed from existing base metrics. 

We have introduced the following two additional configuration parameters for Calc Metrics:
 

• Script: You can select the script that you want to associate with the selected Calc Metric, meaning, the selected script will get mapped to the selected Calc Metric.

• Configuration: You can add or edit the JSON configuration for the selected Calc Metric.

Data Modeler - Using Data Dictionary UI to select the NLP Datasource for a dimension

Now from the datasources defined in the data model, you can choose a datasource for a particular dimension. The chosen datasource is then used for NLP updates for the selected dimension(s).

To do this, 
 

1.  Go to the Admin console > Data Modeler > Pipeline Manager > Data Dictionary > Dimension tab.

This datasource (Market access) is used for NLP updates for dimension 'Product'.

Performance Monitor - Generate audit logs on user creation , password change and more

Now, WhizAI captures the following activities in the Admin console > Audit Logs page:
 

Users & Security - Enhancements in user onboarding

Now as an Admin user, when you add users to WhizAI, they are notified by an email. Also when adding a new user, you can select a predefined email template.

You can add and define a template from: Admin console > Users & Security > Email Templates.

Also from the Admin console > Content Manager > Configurations;

You can enable or disable email notification for new user creation. Also you can add Admin email, to which user creation email will be sent.

User interface UI improvements

Please note that following User Interface elements are updated:
 

Earlier, the first data model in the data model dropdown list was auto selected by default. Now, this data model field is empty, and users have to select the data model.

NLP Workbench - Description box for Replacement

Now when you add replacements, you can also write the description for that replacement in the Description box.

Thus, whenever you revisit the Replacement page you can see the description against each replacement and understand the reason for adding the particular replacement.

NLP Workbench - Adding synonyms for metadata

Now, when you add singular synonyms for metadata, the plural form of that synonyms also gets added automatically and vice versa.

To add a synonym for metadata:

Narratives - Viewing card level narratives on pinboards

Now, you can view and / or edit card level narratives on pinboards.

As a board owner or editor, you can edit the narratives. The Edit option is displayed when you hover the cursor on the narrative. If you are a board viewer you can only view the narratives.

To view and / or edit the narratives:
 

The Narratives panel is displayed as shown in the following figure.

Pinboard Level Narratives Layout from the Design Mode is displayed.

The On/Off toggle button and the Edit option for the corresponding narrative gets highlighted.

Click On from the On/Off toggle button to make the narrative available for the card, and then click Save.

Note! You can use Select All option to make narratives for all the cards available or unavailable.

Narratives - Exporting narratives as a part of card

Now, when you export a pinboard to PDF or PPT format, you can export the narratives too along with the cards in the board. 

To export the narratives:

Narratives - Building custom narratives just got better

Introducing new and improved custom narrative template builder.

Now, you can use simplified functions and control buttons which are more user friendly to build custom narratives.

Before:

Now: 

You can build custom narratives using the functions compatible with the intent. 

Narratives - Introducing building blocks to configure condition based narratives

In the previous versions, to configure condition based narratives; you were required to manually add the code blocks for if-else conditions. For more information, refer to the following image.

Before:

Now,  we have introduced the condition builder that has a predefined set of ‘if-else’ and ‘if’ rules and an auto-populated list of functions that are compatible with the intent. For more information, refer to the following image.

Now:

Narratives - Creating narratives around ‘Count’

Now you can use the Count building block (from Card Context) to configure the following types of narratives.

• Total TRx was 10M across 10 brands.
• There are total 10 brands across 5 regions.
• ABC was the best contributing brand (25%) amongst 10 brands.

You can use the count function block as shown below: 

Known Issues

Pinboards, Cards, Responses, and Insights

Administration

Data Connections: 

• For the newly input data in the live connection which is inserted after successful pipeline execution for the connection, the whiz-cards will not be refreshed, and the NLP updates will not happen. The corresponding pipeline needs to be run again for the refresh.
• For Amazon S3 connection: The field "Path inside the bucket" should have the path till the parent directory of the folder where the CSV files are available.

Data Dictionary

• When adding computations, additional computations might get loaded for created data models on UI.
• Unable to resize the columns on the Data Dictionary page to see the complete text under it.