app.py

import streamlit as st
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from dataset_import import load_data
from data_cleaning import clean_data
from data_labelling import label_dataset

st.title("AnalyzeYT Dataset Analysis")

# Load and clean dataset
data = load_data()

if data is not None:
    st.write("Loaded Data Preview:")
    st.write(data.head())
    
    # Clean data
    data = clean_data(data)
    
    st.write("Cleaned Data Preview:")
    st.write(data.head())
    
    # Show data description
    st.write("Data Description:")
    st.write(data.describe())

    # Data visualization options
    st.write("Data Visualization:")
    chart_type = st.selectbox("Select Chart Type", ['Line Chart', 'Bar Chart', 'Histogram'])

    if chart_type == 'Line Chart':
        x_axis = st.selectbox("Select X-axis Column", data.columns)
        y_axis = st.selectbox("Select Y-axis Column", data.columns)
        title = st.text_input("Enter Chart Title", "Line Chart")
        st.line_chart(data[[x_axis, y_axis]])
        st.write(f"Line Chart: {title}")

    elif chart_type == 'Bar Chart':
        x_axis = st.selectbox("Select X-axis Column", data.columns)
        y_axis = st.selectbox("Select Y-axis Column", data.columns)
        title = st.text_input("Enter Chart Title", "Bar Chart")
        st.bar_chart(data[[x_axis, y_axis]])
        st.write(f"Bar Chart: {title}")

    elif chart_type == 'Histogram':
        selected_column = st.selectbox("Select Column for Histogram", data.columns)
        bins = st.slider("Number of Bins", min_value=10, max_value=100, value=30)
        title = st.text_input("Enter Chart Title", "Histogram")
        plt.hist(data[selected_column], bins=bins)
        plt.title(title)
        plt.xlabel(selected_column)
        plt.ylabel('Frequency')
        st.pyplot()

    # Add option for labeling the dataset
    st.write("Labeling Options:")
    label_option = st.radio("Do you want to label your dataset?", ('No', 'Yes'))

    if label_option == 'Yes':
        # Select columns for labeling
        columns_to_label = st.multiselect("Select Columns to Label", data.columns)
        
        # Input new column names for labeled data
        new_column_names = []
        for col in columns_to_label:
            new_name = st.text_input(f"Enter new name for labeled column '{col}':", f"{col}_labeled")
            new_column_names.append(new_name)
        
        output_name = st.text_input("Enter Output File Name", "labeled_dataset.csv")
        if st.button("Run Labeling"):
            labeled_data = label_dataset(data, columns_to_label, new_column_names)
            labeled_data.to_csv(output_name, index=False)
            st.write("Labeling Completed. Download your file below:")
            st.download_button(label="Download Labeled Dataset", data=labeled_data.to_csv(), file_name=output_name, mime='text/csv')