PlotMind 📊

Convert any CSV into meaningful graphs automatically.

PlotMind analyses your CSV, detects column types, recommends the best chart, and generates it — with zero manual configuration. It supports interactive charts via Plotly and static charts via Matplotlib, and can export to PNG, PDF, or HTML.

---

Installation

pip install plotmind

Or install from source:

git clone https://github.com/git-rohan7/plotmind
cd plotmind
pip install -e .

---

Quick Start

Python API

fromplotmindimportload_csv, clean_dataframe, plot, export# 1. Loaddf=load_csv("sales.csv")
# 2. Clean (handles missing values, duplicates, type coercion)df=clean_dataframe(df, verbose=True)
# 3. Plot – auto-detects best chartfig=plot(df)
# 4. Exportexport(fig, "sales_chart.html")
export(fig, "sales_chart.png")

Command-Line Interface

# Auto-detect best chart and display
plotmind sales.csv
# Choose a specific chart type
plotmind sales.csv --chart bar
# Specify axes
plotmind sales.csv --chart scatter --x revenue --y profit
# Export to file without displaying
plotmind sales.csv --chart line --export output.html
# Use matplotlib backend
plotmind sales.csv --backend matplotlib --export chart.png
# Show column info and recommendation
plotmind sales.csv --info

---

Features

Feature	Description
Auto-chart selection	Detects the best chart based on column types
Column type detection	Identifies numerical, categorical, datetime, boolean columns
Data cleaning	Fills missing values, removes duplicates, fixes types
Interactive charts	Plotly backend for zoom/hover/pan
Static charts	Matplotlib backend for PNG/PDF output
Export	PNG, PDF, HTML, SVG
CLI	Full command-line interface

---

Chart Types

PlotMind supports these chart types (auto-selected or manually chosen):

Chart	When auto-selected
histogram	Single numerical column
bar	Categorical + numerical (high cardinality)
pie	Categorical + numerical (≤6 unique categories)
scatter	Two numerical columns
line	Datetime + numerical
heatmap	Many numerical columns (correlation matrix)
box	Box plot for distribution

---

API Reference

load_csv(filepath, **kwargs) → DataFrame

Load a CSV and validate it.

df=load_csv("data.csv")

---

clean_dataframe(df, verbose=False) → DataFrame

Clean: fill missing values, remove duplicates, coerce types.

df=clean_dataframe(df, verbose=True)

---

detect_columns(df) → dict

Return a dict of {column_name: type} for each column. Types: numerical, categorical, datetime, boolean, unknown.

fromplotmindimportdetect_columnstypes=detect_columns(df)
# {'age': 'numerical', 'city': 'categorical', ...}

---

recommend_chart(df, x=None, y=None) → (chart, x_col, y_col)

Recommend the best chart. Returns a tuple.

fromplotmindimportrecommend_chartchart, x, y=recommend_chart(df)
print(chart) # e.g. 'scatter'

---

plot(df, chart=None, x=None, y=None, title=None, backend='plotly', show=True) → Figure

Generate a chart. Returns the figure object.

fig=plot(df, chart="bar", x="city", y="sales", title="Sales by City")
fig=plot(df, backend="matplotlib", show=False) # static, no display

---

export(fig, path, fmt=None) → str

Export the figure. Format is inferred from the file extension.

export(fig, "chart.png") # PNG (requires kaleido for Plotly)export(fig, "chart.pdf") # PDFexport(fig, "chart.html") # Interactive HTML (Plotly) or embedded HTML (Matplotlib)export(fig, "chart.svg") # SVG

---

Utils

fromplotmind.utilsimportpreview, column_stats, filter_columnspreview(df) # Print headcolumn_stats(df) # Summary DataFrame with null counts, unique values, etc.filter_columns(df, types=["number"]) # Get only numeric column names

---

Testing

pip install plotmind[dev]
pytest tests/ -v

---

Requirements

• Python ≥ 3.8
• pandas ≥ 1.3
• plotly ≥ 5.0
• matplotlib ≥ 3.4
• kaleido ≥ 0.2.1 (for Plotly static image export)

Optional:

• seaborn (for nicer heatmaps with pip install plotmind[seaborn])

---

License

MIT