Data Dashboard with Pandas and Streamlit

Introduction

Data is everywhere — spreadsheets, CSV exports, survey results, sales records, and API responses. The ability to quickly explore, filter, and visualise a dataset is one of the most valuable skills a developer or analyst can have.

In this tutorial you will build a fully interactive Data Dashboard using Pandas and Streamlit. Users can upload any CSV file and immediately get column statistics, apply filters, sort data, and view bar charts, line charts, and scatter plots — all without writing a single line of code after the app is deployed.

This project pairs well with the File Organizer tutorial for working with local files, and with Sentiment Analysis for exploring text data.

What You Will Build

CSV Upload — drag-and-drop any CSV file

Data Preview — first N rows with column types

Summary Statistics — count, mean, min, max, std for numeric columns

Column Filter — select which columns to display

Value Filter — filter rows by any column value

Sorting — sort by any column, ascending or descending

Charts — bar, line, and scatter plots

Download — export the filtered dataset as a new CSV

Prerequisites

Python 3.8+ — python.org

Libraries:

code

pip install streamlit pandas

Package	Version	Purpose
`streamlit`	≥ 1.32	Web app framework
`pandas`	≥ 2.0	Data loading, filtering, aggregation
`io`	built-in	In-memory CSV export

Step 1: Page Setup and File Upload

Create dashboard.py:

code

import io
import pandas as pd
import streamlit as st

st.set_page_config(
    page_title="Data Dashboard",
    page_icon="📊",
    layout="wide"
)

st.title("📊 Interactive Data Dashboard")
st.write("Upload a CSV file to explore, filter, and visualise your data instantly.")

uploaded_file = st.file_uploader(
    "Choose a CSV file",
    type=["csv"],
    help="The file should have a header row. Max recommended size: 50 MB."
)

Step 2: Load and Cache the Data

Use st.cache_data to avoid reloading the file on every Streamlit rerun:

code

@st.cache_data
def load_data(file) -> pd.DataFrame:
    return pd.read_csv(file)

if uploaded_file is not None:
    df = load_data(uploaded_file)
    st.success(f"✅ Loaded **{len(df):,} rows** and **{len(df.columns)} columns**")
else:
    st.info("👆 Upload a CSV file above to get started.")
    st.stop()

st.cache_data stores the parsed DataFrame in memory. The function only runs again if the uploaded file changes, which keeps the app snappy even for large files.

Step 3: Data Overview

Show the shape, column types, and a configurable number of preview rows:

code

with st.expander("🔍 Data Overview", expanded=True):
    col1, col2, col3 = st.columns(3)
    col1.metric("Rows", f"{len(df):,}")
    col2.metric("Columns", len(df.columns))
    col3.metric("Missing Values", int(df.isnull().sum().sum()))

    preview_rows = st.slider("Rows to preview", 5, 50, 10)
    st.dataframe(df.head(preview_rows), use_container_width=True)

    st.write("**Column Types:**")
    col_types = pd.DataFrame({
        "Column": df.columns,
        "Type": df.dtypes.astype(str).values,
        "Non-Null Count": df.notnull().sum().values,
        "Null Count": df.isnull().sum().values,
    })
    st.dataframe(col_types, use_container_width=True, hide_index=True)

Step 4: Summary Statistics

code

with st.expander("📈 Summary Statistics"):
    numeric_cols = df.select_dtypes(include="number").columns.tolist()
    if numeric_cols:
        st.dataframe(
            df[numeric_cols].describe().round(2),
            use_container_width=True
        )
    else:
        st.info("No numeric columns found in this dataset.")

df.describe() returns count, mean, std, min, 25th/50th/75th percentile, and max for every numeric column — a one-liner that saves hours of manual calculation.

Step 5: Column and Value Filters

Let users narrow down the data using the sidebar:

code

st.sidebar.header("🔧 Filters")

# Column selector
all_cols = df.columns.tolist()
selected_cols = st.sidebar.multiselect(
    "Show columns:", all_cols, default=all_cols[:min(6, len(all_cols))]
)

# Value filter for categorical columns
cat_cols = df.select_dtypes(include="object").columns.tolist()
filtered_df = df.copy()

for col in cat_cols[:3]:  # limit to first 3 categorical columns
    unique_vals = df[col].dropna().unique().tolist()
    if len(unique_vals) <= 30:  # only show filter for low-cardinality columns
        chosen = st.sidebar.multiselect(
            f"Filter by {col}:", unique_vals, default=unique_vals
        )
        filtered_df = filtered_df[filtered_df[col].isin(chosen)]

# Numeric range filters
for col in df.select_dtypes(include="number").columns[:2]:
    min_val = float(df[col].min())
    max_val = float(df[col].max())
    chosen_range = st.sidebar.slider(
        f"{col} range:",
        min_value=min_val,
        max_value=max_val,
        value=(min_val, max_val)
    )
    filtered_df = filtered_df[
        filtered_df[col].between(chosen_range[0], chosen_range[1])
    ]

Step 6: Sortable Filtered Table

code

st.subheader("📋 Filtered Data")

col_sort, col_order = st.columns(2)
with col_sort:
    sort_by = st.selectbox("Sort by:", selected_cols)
with col_order:
    ascending = st.radio("Order:", ["Ascending", "Descending"]) == "Ascending"

display_df = (
    filtered_df[selected_cols]
    .sort_values(by=sort_by, ascending=ascending)
    .reset_index(drop=True)
)

st.info(f"Showing **{len(display_df):,}** of **{len(df):,}** rows")
st.dataframe(display_df, use_container_width=True)

Step 7: Charts

Provide three chart types — bar, line, and scatter:

code

st.divider()
st.subheader("📊 Visualisations")

numeric_cols = filtered_df.select_dtypes(include="number").columns.tolist()
all_cols_list = filtered_df.columns.tolist()

if not numeric_cols:
    st.warning("No numeric columns available for charting.")
else:
    chart_type = st.radio(
        "Chart Type:", ["Bar Chart", "Line Chart", "Scatter Plot"],
        horizontal=True
    )

    if chart_type in ["Bar Chart", "Line Chart"]:
        x_col = st.selectbox("X axis:", all_cols_list, key="x")
        y_col = st.selectbox("Y axis:", numeric_cols, key="y")
        chart_data = filtered_df[[x_col, y_col]].dropna()
        if chart_type == "Bar Chart":
            st.bar_chart(chart_data.set_index(x_col))
        else:
            st.line_chart(chart_data.set_index(x_col))

    elif chart_type == "Scatter Plot":
        x_col = st.selectbox("X axis:", numeric_cols, key="sx")
        y_col = st.selectbox("Y axis:", numeric_cols, key="sy")
        st.scatter_chart(
            filtered_df[[x_col, y_col]].dropna(),
            x=x_col, y=y_col
        )

Step 8: Download Filtered Data

code

st.divider()
st.subheader("⬇️ Export")

buffer = io.StringIO()
display_df.to_csv(buffer, index=False)
st.download_button(
    label="⬇️ Download Filtered CSV",
    data=buffer.getvalue(),
    file_name="filtered_data.csv",
    mime="text/csv",
    use_container_width=True,
)

Run the App

code

streamlit run dashboard.py

Pandas Cheat Sheet for This Project

Task	Pandas Code
Load CSV	`pd.read_csv("file.csv")`
First N rows	`df.head(N)`
Column data types	`df.dtypes`
Count missing values	`df.isnull().sum()`
Select numeric columns	`df.select_dtypes(include="number")`
Filter rows by value	`df[df["col"] == value]`
Filter rows by range	`df[df["col"].between(a, b)]`
Sort rows	`df.sort_values(by="col", ascending=True)`
Summary statistics	`df.describe()`
Select columns	`df[["col1", "col2"]]`
Reset index	`df.reset_index(drop=True)`
Export to CSV string	`df.to_csv(index=False)`

Streamlit Data Display Components

Component	Best For
`st.dataframe()`	Interactive, sortable, resizable table
`st.table()`	Static, read-only table
`st.bar_chart()`	Quick bar charts from a DataFrame
`st.line_chart()`	Time series and trend lines
`st.scatter_chart()`	Correlation and distribution plots
`st.metric()`	Single KPI numbers with delta
`st.json()`	Display raw JSON data

Handling Common CSV Issues

Problem	Fix
Wrong encoding	`pd.read_csv(file, encoding="latin-1")`
Semicolon delimiter	`pd.read_csv(file, sep=";")`
No header row	`pd.read_csv(file, header=None)`
Mixed numeric/text in a column	`pd.to_numeric(df["col"], errors="coerce")`
Date columns read as strings	`pd.to_datetime(df["date_col"])`
Memory error on large file	`pd.read_csv(file, chunksize=10000)`

What to Try Next

Add correlation heatmap using st.pyplot() with seaborn.heatmap(df.corr()).

Support Excel files (.xlsx) by adding openpyxl and using pd.read_excel().

Add group-by aggregation — group by a categorical column and show mean/sum per group.

Show outlier detection — flag rows where any numeric value is more than 3 standard deviations from the mean.

Cache results to SQLite using the built-in sqlite3 module.

Sentiment Analysis App — Another data-focused Streamlit app, this time analysing the polarity of text rather than numeric data.

File Organizer — Uses Python's pathlib and shutil to work with files on disk, a great complement to CSV-based data work.

Recipe Finder — Shows how to work with structured JSON data from an API, similar to working with tabular CSV data.

Conclusion

You have built a fully featured interactive data dashboard in Python using Pandas and Streamlit. The app handles CSV uploads, shows summary statistics, provides multi-column filtering, supports three chart types, and exports the filtered result — all with minimal code and zero frontend experience required.

Pandas and Streamlit together are one of the most productive combinations in the Python ecosystem for rapid data exploration and sharing.

Resources:

Pandas Documentation

Pandas Cheat Sheet (PDF)

Streamlit Charts Documentation

Streamlit st.cache_data

Streamlit st.dataframe