Streamlit — Survey Analysis

Summary + Chatbot + Storytelling + Verbatim Analysis

Features

Chatbot

• Deterministic descriptive analyses:
  • Counts
  • Top-K distribution
  • Unique values
  • Search "contain ..."
  • Percentage of a value
  • Numerical statistics
• RAG / LLM for qualitative questions

Main functionalities

• Global-summary:
  • Detailed storytelling generated after an exhaustive analysis of responses and respondent characteristics
  • Documentation of analytical conclusions, key insights, and identified obstacles according to the survey theme (points of interest, blocking factors, etc.)
• Mini-summary:
  • Automatic analysis by question
  • Wordcloud for text columns
  • Sentiment pie chart (hidden if no results)
• Exact count: "How many answered … " across the entire column
• Automatic suggestion of columns to ignore (IDs, names, emails...) with apply / reset option
• Verbatim analysis:
  • Descending sort of labels
  • Filters by labels / sentiments
  • Mood indicators
  • Sentiment pie chart
  • Pagination (20 responses per page)
  • Persistent charts

Requirements

Install the required dependencies:

pip install streamlit pandas numpy scikit-learn unidecode openai openpyxl altair wordcloud markdown