app.py - master - Student_Sw/my_app

app.py
188 строк7 KB
swetiksem234
Create: app.py, items.csv, orders.csv, users.csv
24 июл 2026, 22:03
24 июл 2026, 22:034792409
import streamlit as st
import pandas as pd
import matplotlib.pyplot as plt
from pandas.api.types import is_datetime64_any_dtype
from pandas.api.types import is_numeric_dtype

@st.cache_data
def load_data():
    #... Загрузка данных ...
    url_orders = "orders.csv"
    url_users = "users.csv"
    url_items = "items.csv"
    
    data_orders = pd.read_csv(url_orders)
    data_users = pd.read_csv(url_users)
    data_items = pd.read_csv(url_items)
    
    data_orders = data_orders.drop_duplicates()
    data_users = data_users.drop_duplicates()
    data_items = data_items.drop_duplicates()
    
    # ... обработка данных ...
    # объединение
    data = (
        data_orders
        .merge(
            data_users,
            on='user_id',
            how='left'
        )
        .merge(
            data_items,
            on='item_id',
            how='left'
        )
    )
    nulls = data.isnull().sum()
    print(nulls)
    types = {"order_date": "datetime", "registration_date":"datetime", "price_per_unit": "float64", "quantity": "int64", "base_price": "float64"}
    print(data.dtypes)
    for col in data.columns:
        if (col in types and data[col].dtype != types[col]):
            if (types[col] == "float64" or types[col] == "int64" and is_numeric_dtype(data[col]) == False):
                data[col] = pd.to_numeric(data[col], errors='coerce')
            elif (types[col] == "datetime" and is_datetime64_any_dtype(data[col]) == False):  
                data[col] = pd.to_datetime(data[col], errors='coerce') 
                
    print(data.dtypes)
    base_values = {"category": "Unknown", "user_name": "Unknown"}
    del_values = ["registration_date", "price_per_unit", "quantity"]
    data = data.dropna(subset=del_values)
    nulls = data.isna().sum()
    print(nulls)
    for col in nulls.index:
        if (nulls[col] > 0 and col in base_values):
            data[col] = df[col].fillna(base_values[col])
                
    return data


# В основном коде приложения
st.set_page_config(layout="wide")
st.title("Анализ продаж")

df = load_data()

tab1, tab2, tab3, tab4 = st.tabs(["Сырые данные", "Ключевые показатели", "Визуализация", "Аналитические выводы"])

with tab1:
    
    st.header("Сырые данные")
    
    with st.sidebar:
        st.header("Фильтры")

        #df_visib = df.copy()
        #df_visib['order_date'] = df['order_date'].dt.strftime('%d-%m-%Y')
        #df_visib['registration_date'] = df['registration_date'].dt.strftime('%d-%m-%Y')
        dates = sorted(list(df['order_date'].unique()))
        date = st.multiselect('Дата заказа:', dates, default=dates)
        
        categories = sorted(list(df['category'].unique()))#.sort_values()
        category = st.multiselect('Категория:', categories, default=categories)       
        filtered_df = df[(df['order_date'].isin(date)) & (df['category'].isin(category))]

        
    st.dataframe(filtered_df)
    

with tab2:
    st.header("Ключевые показатели")
    count_orders = len(df)
    total_revenue = (df["quantity"]*df["price_per_unit"]).sum()
    users = df["user_id"].drop_duplicates()
    count_users_uniq = len(users)
    
    
    col1, col2, col3, col4 = st.columns(4)
    
    with col1:
        st.metric("Общее количество заказов", f"{count_orders:}")
        
    with col2:
        st.metric("Общая выручка", f"{total_revenue:,.0f}₽")  
        
    with col3:
        st.metric("Количество уникальных пользователей", f"{count_users_uniq:}")    
        
    with col4:
        st.metric("Средний чек", f"{total_revenue/count_orders:,.0f}₽")       
    
   # st.dataframe(df)

with tab3:
    st.header("Визуализация")
    col5, col6 = st.columns(2)
    df["revenue"] = df["quantity"]*df["price_per_unit"]
    
    item_revenue = (
        df
        .groupby('item_name')
        .agg(
            total_revenue=('revenue', 'sum'),
        )
        .sort_values('total_revenue', ascending=False)
        .head(10)
        .reset_index()  
    ) 
    
    with col5:
        st.subheader("Топ 10 товаров по выручке")   
        fig_bar, axs = plt.subplots(figsize=(10, 6))
        bars = axs.barh(item_revenue['item_name'], item_revenue['total_revenue'])
        axs.bar_label(bars, label_type='edge', color='black')
        st.pyplot(fig_bar)
    
    category_revenue = (
        df
        .groupby('category')
        .agg(
            total_revenue=('revenue', 'sum'),
        )
        .sort_values('category', ascending=False)
        .reset_index()  
    )
    category_revenue["total_revenue_pr"] = category_revenue["total_revenue"] / total_revenue * 100
    with col6:
        fig_pie, ax = plt.subplots(figsize=(10, 6))
        pie = ax.pie(category_revenue["total_revenue_pr"], labels=round(category_revenue["total_revenue_pr"], 2))
        #ax.pie_label(pie, '{frac:.1%}g)', textprops=dict(color="w", size=8, weight="bold"))
        ax.legend(pie.wedges, category_revenue["category"],
            loc="center left",
            bbox_to_anchor=(1, 0, 0.5, 1))
        st.subheader("Выручка по категориям")
        st.pyplot(fig_pie)
    
    df["weekday"] = df["order_date"].dt.weekday
    weekday_revenue = (
            df
            .groupby('weekday')
            .agg(
                count_orders=('order_id', 'sum'),
            )
            .sort_values('weekday', ascending=False)
            .reset_index()  
        )        
    
    st.subheader("Зависимость количества заказов от дня недели")   
    fig_weekday, ax_weekday = plt.subplots(figsize=(10, 6))
    bar_w = ax_weekday.bar(weekday_revenue['weekday'], weekday_revenue['count_orders'])
    ax_weekday.set_xticklabels([' ', 'Понедельник', 'Вторник', 'Среда', 'Четверг', 'Пятница', 'Суббота', 'Воскресенье'])  # Новые метки
    ax_weekday.bar_label(bar_w, label_type='edge', color='black')
    st.pyplot(fig_weekday)   
        
  #  st.dataframe(df)
    
with tab4:
    st.header("Аналитические выводы")
    
    st.markdown(f"Основная выручка приходится на категорию \"{category_revenue.sort_values("total_revenue", ascending=False)['category'].head(1)[0]}\"")    
    days = {0: "Понедельник", 1: "Вторник", 2:"Среда", 3:"Четверг", 4: "Пятница", 5:"Суббота", 6:"Воскресенье"}
        
    weekday_revenue['weekday'] = weekday_revenue['weekday'].map(days)
    days_ = weekday_revenue.sort_values('count_orders', ascending=False)['weekday'].head(2).reset_index()['weekday']
    print(days_)
    st.markdown(f"Пик заказов наблюдается в день: {", ".join(days_)}")   
    
    st.markdown(f"Товар \"{item_revenue.sort_values("item_name", ascending=False)['item_name'].head(1).reset_index()["item_name"][0]}\" является лидером продаж")