Update chatbot with latest context

Format a template with provided values Args: template: Template string **kwargs: Values to insert

Generate text using OpenAI API

Generate metadata about the loaded dataset

Get context about specific KPIs

Add a conversation turn

Analyze a specific KPI's performance Args: kpi_name: Name of the KPI to analyze Returns:

Get specific KPI by name Args: kpi_name: Name of the KPI Returns: KPI dictionar

Classify cardinality of categorical variable

Classify correlation strength

Classify distribution type

Classify missing value severity

Classify outlier severity

Classify trend severity

Classify trend strength based on R-squared

Generate insights about categorical variables

Generate insights about correlations

Generate insights about distributions

Generate insights about missing values

Generate insights about outliers

Generate response using LLM Args: question: User question context: Retrieved context

Generate insights about trends

Get insights related to anomalies

Get context for comparison queries

Get correlation context for explanations

Get insights that help explain phenomena

Get general dataset statistics

Get context about outliers and anomalies

Get top KPIs from dashboard

Get context related to trends

Get insights related to trends

Tab 2: Run EDA and view insights

Analyze categorical variables

Analyze correlations between numeric variables

Analyze distributions of numeric variables

Analyze missing values in the dataset

Main chat interface - answer user questions Args: question: User's question Returns:

Classify the intent of a user query Args: query: User's natural language query Returns:

Clear conversation history

Compute aggregated metrics for dashboard creation

Create LLM client based on provider Args: provider: LLM provider (openai, anthropic, mock) api_key:

Tab 3: Dynamic dashboards from dataset

Detect outliers using IQR and Z-score methods

Detect trends in time series or sequential data

Generate natural language explanation of a KPI Args: kpi_name: Name of the KPI Returns:

Extract entities from query (column names, values, etc.) Args: query: User query Returns:

Format recent history for LLM context

Generate all insights from EDA results Returns: List of insight dictionaries

Generate clarification question for ambiguous queries Args: query: User query intent: Classified int

Get all KPIs Returns: List of KPI dictionaries

Get comprehensive dashboard overview Returns: Dashboard overview dictionary

Get descriptive statistics for numerical columns

Get recent conversation turns

Initialize Streamlit session state

Load analysis context for the chatbot Args: eda_results: EDA analysis results insights: Generated in

Load Power BI metadata from JSON file Args: metadata_path: Path to metadata JSON file Retur

Main application

Determine if query needs clarification Args: query: User query intent: Classified intent

Reset conversation history

Retrieve relevant context for a query Args: query: User query intent: Classified intent

Run complete EDA analysis Returns: Dictionary containing all analysis results

Chatbot in sidebar - always visible

Update the available context

Tab 1: Upload data and Power BI metadata

Generate comprehensive dashboard explanation Returns: Dashboard explanation string

Explain a specific insight in detail Args: insight: Insight dictionary Returns:

Explain a specific KPI from Power BI dashboard Args: kpi_name: Name of the KPI Returns:

Export dashboard context in LLM-friendly format Returns: Formatted context string

Find KPIs related to a given KPI Args: kpi_name: Name of the KPI Returns: List

Format retrieved context for LLM consumption Args: context: Retrieved context dictionary Re

Format insights as a text report Returns: Formatted string report

Generate text from prompt

Generate mock response based on prompt keywords Args: prompt: User prompt system_prompt: System inst

Generate text using Anthropic API

Get configuration for a specific section or all configuration Args: section: Configuration section name (llm, analysis, powerbi,

Get insights filtered by category Args: category: Category to filter by Returns:

Get a text summary of all KPIs Returns: Formatted string summary

Get dataset metadata Returns: Metadata dictionary

Get a sample of the data Args: n: Number of rows to return Returns: Sample Data

Get a template by name Args: name: Template name Returns: Template string

Get top N insights by priority Args: n: Number of insights to return Returns: L

Load data from CSV or Excel file Args: file_path: Path to the data file Returns:

Generate dashboard summary Returns: Dashboard summary