Plottle › GUI Guide

GUI Tutorial — Plottle

This guide walks through the Streamlit app page by page, then shows a complete end-to-end workflow: upload data → analyze → plot → export.

Launching the App

# Activate your virtual environment first, then:
streamlit run plottle/Home.py

Open http://localhost:8501 in your browser. The sidebar on the left lists all pages.

Page 1 — Data Upload

Purpose: Load one or more data files into the session. All subsequent pages work with the data loaded here.

Supported formats

Extension	Returns
`.csv`, `.xlsx`, `.tsv`, `.json`, `.parquet`	pandas DataFrame
`.npy`, `.npz`	NumPy array
`.pkl`	Any Python object

Steps

Click Browse files (or drag-and-drop) to upload your file.
The page shows a preview: shape, column types, and summary statistics.
Upload additional files — each becomes a named dataset in the sidebar selector.
Select the active dataset from the dropdown at the top of every page.

Tips

Large files (> 50 MB) may be slow to preview but will work.
.npz files load all arrays; select which array to use in subsequent pages.

Page 2 — Quick Plot

Purpose: Generate any of the 13 supported plot types with point-and-click controls.

Plot types available

Category	Types
Matplotlib (static)	Line, Scatter, Histogram, Heatmap, Contour
Seaborn (statistical)	Distribution, Box / Violin, Regression
Plotly (interactive)	Histogram, Scatter, Line, Heatmap, 3D Surface

Steps

Select a plot type from the dropdown.
Choose your X column and Y column (or the data array for histograms).
Adjust style options: title, axis labels, color palette, line width, etc.
Click Generate Plot.
Plotly charts are interactive — zoom, pan, and hover for values.
Expand Export plot & data to download the figure as PNG, SVG, PDF, or HTML.

Tips

Switching plot type resets the column selectors — pick the type first.
For multi-series line plots, select multiple Y columns with Ctrl+click.
Plotly HTML exports are self-contained and can be shared as a single file.

Page 3 — Analysis Tools

Purpose: Run quantitative analysis on the active dataset. Five tabs cover the most common scientific workflows.

Tab A — Descriptive Statistics

Displays mean, median, std, variance, min, max, Q1, Q3, IQR, and range for every numeric column. Also runs a Shapiro-Wilk normality test with pass/fail indicator.

Tab B — Distribution Fitting

Fit a theoretical distribution (Normal, Exponential, Log-normal, etc.) to a numeric column. Shows fitted parameters, KS-test statistic, and an overlay plot.

Tab C — Curve Fitting

Fit a model to X/Y data:

Model	Parameters
Linear	slope, intercept, R²
Polynomial	degree 1–10, coefficients, R²
Exponential	a, b (y = a·exp(b·x)), R²
Custom	Enter any Python expression using `x`, e.g., `a * np.sin(b * x) + c`

Results show fitted parameters and a residuals plot. Click Copy Parameters to paste values into a script.

Tab D — Optimization

Find a function minimum or root:

Minimize: Enter a Python expression in x (e.g., (x - 3)**2 + 1); set bounds and initial guess.
Find root: Enter an expression; set a bracket [a, b] that contains a sign change.

Tab E — Linear Algebra

Upload or enter a matrix, then compute:

Eigenvalues and eigenvectors
Solve a linear system A·x = b
Matrix decomposition (QR, SVD, Cholesky)

Page 4 — Multi-Plot Dashboard

Purpose: Display up to six independent plots side-by-side in a configurable grid.

Steps

Choose a grid layout (1×1 up to 2×3).
For each cell, select a dataset, plot type, and columns independently.
Click Generate Dashboard — all plots render simultaneously.
Expand any cell to enlarge it; click its Download button to save.

Tips

Each cell is independent — you can mix Matplotlib and Plotly in the same dashboard.
Use the 2×3 layout for a six-panel publication figure.

Page 5 — Advanced Plotting

Purpose: Specialized visualizations not available on the Quick Plot page.

Options

Feature	Description
Correlation Heatmap	Pearson correlation matrix for all numeric columns
Overlaid Distributions	KDE / ECDF overlaid across groups (color-coded by a categorical column)
Grouped Categorical	Box, violin, or strip plot grouped by a categorical column
3D Scatter	Interactive Plotly 3D scatter with color mapping
HTML Export	Save any Plotly figure as a self-contained HTML file

Page 6 — Settings

Purpose: Persist plot defaults and create named style presets.

Configurable defaults

Color palette — choose from six palettes including two color-blind-safe options
Figure size — width × height in inches
DPI — default resolution for saved figures
Grid — on/off
Font size — base font for all text
Line width — default line thickness
Legend — show/hide

Presets

Click Save as preset to name and store the current defaults. Switch between presets from the dropdown. Settings are written to config.json in the project root and persist across app restarts.

Page 7 — Export Results

Purpose: Download figures, datasets, and analysis results generated during the session.

Plot history

Every plot generated on Page 2 or Page 5 is recorded in the session history. For each:

Choose a format: PNG 150 dpi, PNG 300 dpi, SVG, PDF, or HTML (Plotly)
Click Download to save the file

Dataset export

Export the currently loaded dataset as CSV, JSON, NumPy (.npy), or pickle (.pkl).

Analysis results

Export any curve-fit or statistics result from Page 3 as a structured JSON file.

Session save / load

Save session — writes all loaded datasets, plot history, and analysis results to a single .json file you can download.
Load session — drag-and-drop a previously saved session file to restore the full state.

Complete Workflow Example

Goal: Upload a calibration dataset, fit a Beer-Lambert curve, export the fit result.

Page 1: Upload calibration.csv (columns: concentration, absorbance).
Page 3 → Curve Fitting: Select concentration (X) and absorbance (Y); choose Linear; click Fit. Note the slope (molar absorptivity · path length) and R².
Page 2: Select Scatter plot; X = concentration, Y = absorbance; title = "Beer-Lambert Calibration"; click Generate Plot.
Page 7: Download the scatter plot as PNG 300 dpi. Download the curve-fit result as JSON.

Total time: under two minutes with no code written.