pages/EDA .py

import streamlit as st
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Function to generate automatic insights for univariate analysis
def generate_univariate_insights(data, column):
    mean_val = data[column].mean()
    median_val = data[column].median()
    std_val = data[column].std()
    min_val = data[column].min()
    max_val = data[column].max()

    insights = f"""
    - The mean value of '{column}' is {mean_val:.2f}.
    - The median value is {median_val:.2f}, indicating the central tendency of the data.
    - The standard deviation is {std_val:.2f}, suggesting the spread of the values.
    - The minimum value observed is {min_val}, and the maximum value is {max_val}.
    """
    return insights

# Function to generate automatic insights for bivariate analysis (scatter plot)
def generate_bivariate_insights(data, col1, col2):
    correlation = data[col1].corr(data[col2])
    
    insights = f"""
    - The correlation between '{col1}' and '{col2}' is {correlation:.2f}.
    - A correlation close to 1 indicates a strong positive relationship, while a correlation close to -1 indicates a strong negative relationship.
    - A correlation near 0 suggests no linear relationship between the variables.
    """
    return insights

# Function to generate automatic insights for multivariate analysis (pairplot)
def generate_multivariate_insights(data, columns):
    correlations = data[columns].corr()
    insights = f"""
    - The pairplot shows the relationships between the selected numeric variables: {', '.join(columns)}.
    - The diagonal displays the distributions of each variable.
    - Strong correlations (positive or negative) can be seen in the scatter plots between some variables.
    """
    return insights

# Introduction to EDA
st.markdown("""
# Exploratory Data Analysis (EDA)
Exploratory Data Analysis (EDA) is an essential step in the data analysis process. It involves:
- **Understanding the Structure**: By examining the dataset’s statistics and structure, we can identify patterns, trends, and potential issues.
- **Visualizing Distributions**: Histograms and boxplots give insight into the distribution of data, the spread of numerical values, and the presence of any outliers.
- **Finding Relationships**: Through scatter plots and correlation matrices, we can identify relationships between two or more variables, which helps in building predictive models.

EDA helps in:
- Cleaning the dataset by handling missing values, detecting outliers, and fixing errors.
- Gaining insights that can inform further analysis or modeling steps.
""")

# File uploader for dataset
uploaded_file = st.file_uploader("Upload your dataset (CSV format):", type=["csv"])

if uploaded_file is not None:
    # Read and display the dataset
    data = pd.read_csv(uploaded_file)
    st.write("### Uploaded Dataset:")
    st.dataframe(data)

    # Dataset Overview
    st.write("### Dataset Overview:")
    st.write(data.describe())

    # Missing values in the dataset
    st.write("### Missing Values:")
    st.write(data.isnull().sum())

    # Correlation matrix for numerical columns
    st.write("### Correlation Matrix:")
    numeric_columns = data.select_dtypes(include=['float64', 'int64']).columns
    if len(numeric_columns) > 1:
        st.write(data[numeric_columns].corr())

        st.write("Heatmap of Correlation Matrix:")
        fig, ax = plt.subplots(figsize=(10, 8))
        sns.heatmap(data[numeric_columns].corr(), annot=True, cmap='coolwarm', ax=ax)
        st.pyplot(fig)

    # Univariate Plots (For a single column)
    st.write("### Univariate Analysis: Distribution of Columns")
    selected_numeric_column = st.selectbox("Select a Numeric Column for Univariate Analysis", numeric_columns)
    
    # Histogram for univariate distribution
    st.write(f"Histogram for '{selected_numeric_column}':")
    fig, ax = plt.subplots()
    sns.histplot(data[selected_numeric_column], kde=True, ax=ax)
    st.pyplot(fig)

    # Display automatic insights for univariate analysis
    univariate_insights = generate_univariate_insights(data, selected_numeric_column)
    st.write("### Insights:")
    st.write(univariate_insights)

    # Boxplot for univariate distribution
    st.write(f"Boxplot for '{selected_numeric_column}':")
    fig, ax = plt.subplots()
    sns.boxplot(x=data[selected_numeric_column], ax=ax)
    st.pyplot(fig)

    # Bivariate Plots (For two columns)
    st.write("### Bivariate Analysis: Relationships between Two Variables")
    selected_bivariate_columns = st.multiselect(
        "Select Two Columns for Bivariate Analysis", 
        options=numeric_columns, 
        default=numeric_columns[:2]
    )
    
    if len(selected_bivariate_columns) == 2:
        st.write(f"Scatter Plot between '{selected_bivariate_columns[0]}' and '{selected_bivariate_columns[1]}':")
        fig, ax = plt.subplots()
        sns.scatterplot(x=data[selected_bivariate_columns[0]], y=data[selected_bivariate_columns[1]], ax=ax)
        st.pyplot(fig)

        # Display automatic insights for bivariate analysis
        bivariate_insights = generate_bivariate_insights(data, selected_bivariate_columns[0], selected_bivariate_columns[1])
        st.write("### Insights:")
        st.write(bivariate_insights)

    # Multivariate Plots (For multiple columns)
    st.write("### Multivariate Analysis: Relationships between Multiple Variables")
    selected_multivariate_columns = st.multiselect(
        "Select Columns for Multivariate Analysis", 
        options=numeric_columns, 
        default=numeric_columns[:3]
    )
    
    if len(selected_multivariate_columns) > 1:
        st.write(f"Pairplot for selected variables: {', '.join(selected_multivariate_columns)}")
        fig, ax = plt.subplots(figsize=(10, 8))
        sns.pairplot(data[selected_multivariate_columns])
        st.pyplot(fig)

        # Display automatic insights for multivariate analysis
        multivariate_insights = generate_multivariate_insights(data, selected_multivariate_columns)
        st.write("### Insights:")
        st.write(multivariate_insights)

    # Categorical vs Numeric (boxplots)
    categorical_columns = data.select_dtypes(include=['object', 'category']).columns
    if len(categorical_columns) > 0:
        selected_cat_column = st.selectbox("Select a Categorical Column for Analysis", categorical_columns)
        
        st.write(f"Boxplot for '{selected_cat_column}' vs Numeric Column:")
        selected_numeric_column_for_cat = st.selectbox("Select a Numeric Column to Plot", numeric_columns)
        fig, ax = plt.subplots()
        sns.boxplot(x=data[selected_cat_column], y=data[selected_numeric_column_for_cat], ax=ax)
        st.pyplot(fig)

        st.write(f"### Insights:")
        st.write(f"Boxplot shows the distribution of '{selected_numeric_column_for_cat}' values for each category in '{selected_cat_column}'. It helps identify if the numerical values differ across categories.")

    # Download the cleaned dataset if needed
    st.markdown("""
    This analysis provides a basic understanding of the dataset. 
    You can now proceed with further analysis or modeling.
    """)
else:
    st.warning("Please upload a dataset to proceed with EDA.")