Анализ данных с помощью ноутбуков Marimo и chDB

export CLICKHOUSE_CLOUD_HOSTNAME=<HOSTNAME>
export CLICKHOUSE_CLOUD_USER=default
export CLICKHOUSE_CLOUD_PASSWORD=your_actual_password

pip install chdb pandas plotly marimo

marimo edit clickhouse_exploration.py

import marimo as mo
import chdb
import pandas as pd
import os
import plotly.express as px
import plotly.graph_objects as go

result = chdb.query("SELECT 'Hello ClickHouse from Marimo!'", "DataFrame")
result

query = f"""
SELECT
    toYear(date) AS year,
    round(avg(price)) AS price,
    bar(price, 0, 1000000, 80)
FROM remoteSecure(
    '{os.environ.get("CLICKHOUSE_CLOUD_HOSTNAME")}',
    'default.pp_complete',
    '{os.environ.get("CLICKHOUSE_CLOUD_USER")}',
    '{os.environ.get("CLICKHOUSE_CLOUD_PASSWORD")}'
)
WHERE town = 'LONDON'
GROUP BY year
ORDER BY year
"""

df = chdb.query(query, "DataFrame")
df.head()

fig = px.line(
    df, 
    x='year', 
    y='price',
    title='Average Property Prices in London Over Time',
    labels={'price': 'Average Price (£)', 'year': 'Year'}
)

fig.update_traces(mode='lines+markers')
fig.update_layout(hovermode='x unified')
fig

town_selector = mo.ui.dropdown(
    options=['LONDON', 'MANCHESTER', 'BIRMINGHAM', 'LEEDS', 'LIVERPOOL'],
    value='LONDON',
    label='Select a town:'
)
town_selector

query_reactive = f"""
SELECT
    toYear(date) AS year,
    round(avg(price)) AS price
FROM remoteSecure(
    '{os.environ.get("CLICKHOUSE_CLOUD_HOSTNAME")}',
    'default.pp_complete',
    '{os.environ.get("CLICKHOUSE_CLOUD_USER")}',
    '{os.environ.get("CLICKHOUSE_CLOUD_PASSWORD")}'
)
WHERE town = '{town_selector.value}'
GROUP BY year
ORDER BY year
"""

df_reactive = chdb.query(query_reactive, "DataFrame")
df_reactive

fig_reactive = px.line(
    df_reactive,
    x='year',
    y='price',
    title=f'Average Property Prices in {town_selector.value} Over Time',
    labels={'price': 'Average Price (£)', 'year': 'Year'}
)

fig_reactive.update_traces(mode='lines+markers')
fig_reactive.update_layout(hovermode='x unified')
fig_reactive

year_slider = mo.ui.slider(
    start=1995,
    stop=2024,
    value=2020,
    step=1,
    label='Select Year:',
    show_value=True
)
year_slider

query_distribution = f"""
SELECT
    price,
    toYear(date) AS year
FROM remoteSecure(
    '{os.environ.get("CLICKHOUSE_CLOUD_HOSTNAME")}',
    'default.pp_complete',
    '{os.environ.get("CLICKHOUSE_CLOUD_USER")}',
    '{os.environ.get("CLICKHOUSE_CLOUD_PASSWORD")}'
)
WHERE town = 'LONDON'
  AND toYear(date) = {year_slider.value}
  AND price > 0
  AND price < 5000000
"""

df_distribution = chdb.query(query_distribution, "DataFrame")

# create an interactive box plot.
fig_box = go.Figure()

fig_box.add_trace(
    go.Box(
        y=df_distribution['price'],
        name=f'London {year_slider.value}',
        boxmean='sd',  # Show mean and standard deviation
        marker_color='lightblue',
        boxpoints='outliers'  # Show outlier points
    )
)

fig_box.update_layout(
    title=f'Distribution of Property Prices in London ({year_slider.value})',
    yaxis=dict(
        title='Price (£)',
        tickformat=',.0f'
    ),
    showlegend=False,
    height=600
)

fig_box