Skip to content
Moira Documentation

Data Analysis

Workflow for data analysis from problem definition to conclusions and visualization. Follows CRISP-DM methodology with validation loops for data quality and analysis completeness.

Start

Section titled “Start”
Terminal window
mcp__moira__start({ workflowId: "data-analysis" })

Process

Section titled “Process”
flowchart LR
    A[Context] --> B[Problem]
    B --> C[Collect Data]
    C --> D[Prepare Data]
    D --> E[Explore/EDA]
    E --> F[Find Insights]
    F --> G[Visualize]
    G --> H[Conclude]

Steps

Section titled “Steps”
StepActionOutput
1. Get ContextCollect business question, context, data sources, constraints, audienceContext document
2. Define ProblemFormulate research question, hypotheses, success criteria, scopeProblem definition
3. Collect DataDownload, study structure, initial quality checkRaw dataset
4. Prepare DataHandle missing values, types, duplicates, outliers, transformationsClean dataset
5. Explore DataDistributions, correlations, patterns, preliminary insightsEDA report
6. Find InsightsTest hypotheses, answer research question, recommendationsKey insights
7. VisualizeCreate charts for key findingsVisualizations
8. ConcludeExecutive summary, findings, recommendations, limitationsFinal report

Features

Section titled “Features”

Validation Loops

Section titled “Validation Loops”
LoopPurposeCriteria
Data qualityVerify data readiness for EDANo critical quality issues
EDA completenessVerify research thoroughnessAll hypotheses addressed

User Approval Gates

Section titled “User Approval Gates”
GateDecision
Problem definitionConfirm research question and scope
ConclusionsApprove final findings and recommendations

Quality Standards

Section titled “Quality Standards”
StandardDescription
ReproducibilityAnalysis can be repeated with same results
ValidityMethods appropriate for data and question
RelevanceFindings address the business question
ClarityResults understandable by audience
ActionabilityRecommendations are practical

Data Preparation Checklist

Section titled “Data Preparation Checklist”
TaskAction
Missing valuesIdentify, understand, handle appropriately
Data typesVerify and correct column types
DuplicatesDetect and remove if necessary
OutliersIdentify, investigate, handle
TransformationsApply needed transformations

Example Node Configuration

Section titled “Example Node Configuration”
{
  "id": "explore-data",
  "type": "agent-directive",
  "directive": "Perform exploratory data analysis. Examine distributions, correlations, and patterns. Document preliminary insights.",
  "completionCondition": "EDA complete with distributions, correlations analyzed and preliminary insights documented",
  "connections": {
    "next": "validate-eda"
  }
}
Section titled “Related”