Analys
Last updated
Last updated
Data Connector skill allows the agent to access external data connectors for advanced reporting and analysis capabilities.
Navigate to "Agent Settings" in the Feedloop AI platform.
Choose the "Skills" tab.
Add a new skill by selecting the "Data Connector" skill.
Choose the type of skill:
Reporting: Configure reporting-type Data Connector skills.
Analysis: Configure analysis-type Data Connector skills.
The reporting-type Data Connector skill enables the agent to generate specific reports based on user queries, enhancing its reporting capabilities.
Skill Configuration
Title: Provide a user-defined title for the reporting-type Data Connector skill.
Input Condition: Specify conditions for activating the skill (e.g., user queries about GWI Score).
Input Reporting Rules: Define specific rules for reporting, such as SQL queries or data manipulation steps.
Adding Connectors
Choose Dataset: Select a dataset for the Data Connector.
Choose Table (Multiple): Choose multiple tables within the selected dataset.
Delete Connector: Remove a connector if needed.
Deleting Connectors
Users can remove connectors under this skill if not needed.
The analysis-type Data Connector skill allows the agent to perform in-depth analysis based on user queries, expanding its analytical capabilities.
Skill Configuration
Input Condition: User queries about the analysis of GWI Score.
Input Reporting Rules: Specific rules for reporting, similar to the reporting type.
Input Analysis Rules: Rules specific to analysis, such as additional SQL queries or data manipulation steps.
Add Connector (Multiple):
Choose Dataset: Select a dataset for the Data Connector.
Choose Table (Multiple): Choose multiple tables within the chosen dataset.
Delete Connector: Remove a connector if needed.
Adding Connectors
Choose Dataset: Select a dataset for the Data Connector.
Choose Table (Multiple): Choose multiple tables within the selected dataset.
Delete Connector: Remove a connector if needed.
Deleting Connectors
Users can remove connectors under this skill if not needed.
Title : GWI_iNDICATOR SCORE
condition : when user ask about GWI Score but not including gwi indicator where gwi score is the main subject of the question
reporting rules :
Use alias for all table
gwi_score = ROUND(SUM(value * category_weight * perspective_weight / 100), 1) group by organization_id
Always select month, year and other required columns
Hospital unit is hospital code
Explanation
Use Alias for All Table:
This rule suggests using aliases for all tables involved in the reporting, providing a clearer and more organized representation of the data.
gwi_score Calculation: = ROUND(SUM(value * category_weight * perspective_weight / 100), 1) GROUP BY organization_id
Calculates the GWI Score based on the given formula. It involves the sum of the product of value, category_weight, and perspective_weight, divided by 100. The result is rounded to one decimal place. The grouping is done by organization_id.
Always Select Month, Year, and Other Required Columns:
Recommends including essential columns like month, year, and other required information in the reporting. This ensures a comprehensive set of data for analysis.
Hospital Unit is Hospital Code
Specifies that in the reporting, the hospital unit should be represented by the hospital code.
Title : GWI_iNDICATOR ANALYSIS
condition : when user asks about the analysis of GWI Score, reasons for GWI Score fluctuations, factors influencing GWI Score, or cause and effect related to GWI Score
reporting rules :
use dataset prefix
gwi_score = SUM(value * category_weight * perspective_weight / 100) MUST group by organization_id
select: category(required), organization_id, archetype(required), column in group by and other columns required for the analysis, and column in user message
required group by: category, organization_id and other columns required for the analysis, and column in user message
if user need to compare some columns, please select the columns
analysis rules :
You must give sample data in the analysis respons
gwi score affected by category and archetype
Explanation Reporting rules
Use Alias for All Table
This rule suggests using aliases for all tables involved in the reporting, providing a clearer and more organized representation of the data
Use Dataset Prefix
This rule suggests using the dataset prefix for clarity and proper identification of data sources.
gwi_score Calculation SQL Query:
Calculates the GWI Score based on the given formula without rounding. The grouping is done by organization_id.
Select Columns for Analysis : select: category(required), organization_id, archetype(required), column in group by and other columns required for the analysis, and column in user message
Recommends selecting specific columns required for the analysis, including category, organization_id, archetype, and other columns mentioned in the user message.
Required GROUP BY Specifies the columns that must be included in the GROUP BY clause for the analysis.
This ensures proper grouping for meaningful insights. Required columns include category, organization_id, and others mentioned in the user message.
Comparison of Columns If the user needs to compare specific columns
they are advised to select those columns for inclusion in the analysis. Analysis Rules
Sample Data Requirement: You must give sample data in the analysis response.
Requires providing sample data in the analysis response, ensuring that users receive illustrative examples related to the analysis.
Factors Affecting GWI Score: gwi score affected by category and archetype
Highlights that in the analysis, attention should be given to the influence of category and archetype on the GWI Score. These analysis rules further refine the expectations for the analysis-type Data Connector skill. They emphasize the inclusion of sample data and specifically point out factors affecting the GWI Score, providing additional context to users.In summary, this analysis-type Data Connector skill activates when users inquire about the deeper analysis of GWI Score, and it applies a set of rules for reporting and analysis to provide comprehensive insights into the data.
Example Result
Multiline Chart
Pie Chart