Dynamic Survey Data Plotting and Tables - A One Stop Streamlit App For Cross Section Data Analysis

Turn raw survey spreadsheets into clean, interactive insights—fast. In this video, we demo a Streamlit + Plotly app for cross-sectional data that auto-detects variable types, runs the right stats, and lets you curate a publication-ready report.

Made by: Noman Arshed
App version shown: 1.0 (code v1.1)
Link: https://www.econistics.com/streamlit....

What you’ll see
• Upload & detect: CSV/XLSX/SPSS (.sav)/Stata (.dta) support, with auto variable typing (nominal/ordinal/continuous) + manual overrides
• Univariate view: Histogram, QQ-normal plot, box plot, descriptive statistics
• Discrete comparisons: Cross-tabs with row/column %, χ² test, optional NaN-group removal, overlay vs faceted histograms, one-way ANOVA with group means/SD/counts, Spearman (when the discrete is ordinal)
• Continuous comparisons: Scatter with optional marginal boxplots, Pearson/Spearman/Kendall correlations
• Ordinal base mirror: If the base is ordinal and you pick a continuous comparator, you get the same grouped hist + ANOVA workflow
• Report & Edit: “Save to Report” freezes any chart; edit titles/axes and type-aware options (e.g., color/size/symbol for scatter; bins/normalization/barmode for hist; points for box). Click “Update chart” to apply changes, then save the edited copy

Chapters
00:00 Intro & goals
00:38 App tour (3 pages)
01:20 Data upload & type overrides
02:25 Univariate plots (Row 1)
03:40 Discrete comparisons (Row 2): cross-tab, χ², % views, ANOVA, Cronbach’s α
06:10 Continuous comparisons (Row 3): scatter, marginals, correlations
08:00 Ordinal base vs continuous (mirrored workflow)
09:15 Report & Edit: type-aware options + “Update chart”
11:10 Export & next steps

Tips & notes
• If you don’t have a continuous variable, the app safely picks any base; downstream errors are trapped.
• Cronbach’s α needs 2+ items (ideally ordered/Likert).
• For SPSS/Stata, the app uses pyreadstat.
• Data stays local to your machine/session.

Feature highlights (quick)
• Auto type detection with editable overrides
• Cross-tabs: counts, row/col %, χ² test
• ANOVA: F, p, η² + group means/SD/counts
• Correlations: Pearson, Spearman, Kendall
• Scatter marginals: optional boxplots on axes
• Overlay vs faceted histograms
• Report builder with frozen charts and per-type controls
• “Update chart” button to apply edits intentionally
• Wrapped text in tables so long labels are readable

Planned enhancements
• Welch ANOVA, Tukey post-hoc
• Export figures (PNG/SVG/PDF)
• Save tables to report (CSV/XLSX)

Credits
Developed by Noman Arshed. Built with Streamlit, Plotly, NumPy/Pandas, SciPy, and pyreadstat.