| user_id | cities          |
|---------|-----------------|
| 1       | [Москва, Сочи]  |
| 2       | [Казань]        |

| user_id | cities  |
|---------|---------|
| 1       | Москва  |
| 1       | Сочи    |
| 2       | Казань  |

cities_popularity = (df.explode('cities')
                     .groupby('cities')['user_id']
                     .count()
                     .sort_values(ascending=False))

target_cities = ['Москва', 'Лондон']
filtered_users = (df.explode('cities')
                  .query('cities in @target_cities')['user_id']
                  .unique())

from itertools import combinations

tags_pairs = (df.explode('cities')
              .groupby('user_id')['cities']
              .apply(lambda x: list(combinations(x, 2)))
              .explode()
              .dropna())

ols_10_logwage = smf.ols('log_wage ~ age + I(age**2) + educ + female + smoke', data=df).fit()
print(ols_10_logwage.summary())

                            OLS Regression Results                            
Ключевые коэффициенты:
age        0.0605 (p=0.109)
age^2     -0.0007 (p=0.077)  ← значим на 10%
female    -0.5787 (p=0.000)  ← значим
educ       0.1646 (p=0.008)  ← значим

import pandas as pd
import numpy as np

dates = pd.date_range('2024-03-01', periods=10, freq='D')
price_data = {
    'Дата': dates,
    'Цена акции, $': [152.3, 148.7, 155.2, 147.1, 159.8, 145.5, 162.3, 158.9, 144.8, 160.1]
}

df = pd.DataFrame(price_data)

print("Динамика цены акции с выделением экстремумов:")
display(df.style
       .highlight_max(subset=['Цена акции, $'], color='lightgreen')
       .highlight_min(subset=['Цена акции, $'], color='#ffcccc'))

from sklearn.metrics import mean_absolute_error, mean_squared_error, mean_absolute_percentage_error
import numpy as np

def calculate_metrics(y_true, y_pred, model_name):
   
    mae = mean_absolute_error(y_true, y_pred)
    mse = mean_squared_error(y_true, y_pred)
    rmse = np.sqrt(mse)
    mape = mean_absolute_percentage_error(y_true, y_pred) * 100
   
    return {
        'Модель': model_name,
        'MAE': mae,
        'MSE': mse,
        'RMSE': rmse,
        'MAPE': mape,
    }

# Пример вычисления метрик для разных моделей
metrics_arima100 = calculate_metrics(y_true, y_pred_arima100, 'ARIMA(1,0,0)')
metrics_arima020 = calculate_metrics(y_true, y_pred_arima020, 'ARIMA(0,2,0)')
metrics_ets = calculate_metrics(y_true, y_pred_ets, 'ETS')

import pandas as pd
metrics_df = pd.DataFrame([metrics_arima100, metrics_arima020, metrics_ets])
metrics_df = metrics_df.set_index('Модель')
print("Метрики качества моделей:")
display(metrics_df.style
        .background_gradient(subset=['MAE', 'MSE', 'RMSE', 'MAPE'], cmap='RdYlGn_r'))

import plotly.graph_objects as go

# Подготовка данных
trip_day_count_dt = (df.groupby('pickup_datetime')['id']
                     .count()
                     .reset_index(name='trip_count')
                     .sort_values(by='pickup_datetime'))
trip_day_count_dt['hour'] = trip_day_count_dt.pickup_datetime.dt.hour
trip_day_count_dt['dayofweek'] = trip_day_count_dt['pickup_datetime'].dt.day_name()

hour_day_trip_count = (trip_day_count_dt
                       .groupby(['hour', 'dayofweek'])['trip_count']
                       .sum()
                       .reset_index(name='trips'))

# Построение графика
fig = go.Figure()
for day in hour_day_trip_count['dayofweek'].unique():
    day_data = hour_day_trip_count[hour_day_trip_count["dayofweek"] == day]
    fig.add_trace(
        go.Scatter(
            x=day_data["hour"], 
            y=day_data["trips"], 
            name=str(day)
        )
    )
    
fig.update_layout(
    title="Динамика поездок по часам суток",
    xaxis_title="Час суток",
    yaxis_title="Количество поездок",
    width=1400, 
    height=600
)
fig.show()