Untitled

visitorsADaily = visitors[visitors['group'] == 'A'][['date', 'visitors']]
visitorsADaily.columns = ['date', 'visitorsPerDateA']

visitorsACummulative = visitorsADaily.apply(
    lambda x: visitorsADaily[visitorsADaily['date'] <= x['date']].agg(
        {'date': 'max', 'visitorsPerDateA': 'sum'}
    ),
    axis=1,
)
visitorsACummulative.columns = ['date', 'visitorsCummulativeA']

visitorsBDaily = visitors[visitors['group'] == 'B'][['date', 'visitors']]
visitorsBDaily.columns = ['date', 'visitorsPerDateB']

visitorsBCummulative = visitorsBDaily.apply(
    lambda x: visitorsBDaily[visitorsBDaily['date'] <= x['date']].agg(
        {'date': 'max', 'visitorsPerDateB': 'sum'}
    ),
    axis=1,
)
visitorsBCummulative.columns = ['date', 'visitorsCummulativeB']

ordersADaily = (
    orders[orders['group'] == 'A'][['date', 'transactionId', 'visitorId', 'revenue']]
    .groupby('date', as_index=False)
    .agg({'transactionId': pd.Series.nunique, 'revenue': 'sum'})
)
ordersADaily.columns = ['date', 'ordersPerDateA', 'revenuePerDateA']

ordersACummulative = ordersADaily.apply(
    lambda x: ordersADaily[ordersADaily['date'] <= x['date']].agg(
        {'date': 'max', 'ordersPerDateA': 'sum', 'revenuePerDateA': 'sum'}
    ),
    axis=1,
).sort_values(by=['date'])
ordersACummulative.columns = [
    'date',
    'ordersCummulativeA',
    'revenueCummulativeA',
]

ordersBDaily = (
    orders[orders['group'] == 'B'][['date', 'transactionId', 'visitorId', 'revenue']]
    .groupby('date', as_index=False)
    .agg({'transactionId': pd.Series.nunique, 'revenue': 'sum'})
)
ordersBDaily.columns = ['date', 'ordersPerDateB', 'revenuePerDateB']

ordersBCummulative = ordersBDaily.apply(
    lambda x: ordersBDaily[ordersBDaily['date'] <= x['date']].agg(
        {'date': 'max', 'ordersPerDateB': 'sum', 'revenuePerDateB': 'sum'}
    ),
    axis=1,
).sort_values(by=['date'])
ordersBCummulative.columns = [
    'date',
    'ordersCummulativeB',
    'revenueCummulativeB',
]

data = (
    ordersADaily.merge(
        ordersBDaily, left_on='date', right_on='date', how='left'
    )
    .merge(ordersACummulative, left_on='date', right_on='date', how='left')
    .merge(ordersBCummulative, left_on='date', right_on='date', how='left')
    .merge(visitorsADaily, left_on='date', right_on='date', how='left')
    .merge(visitorsBDaily, left_on='date', right_on='date', how='left')
    .merge(visitorsACummulative, left_on='date', right_on='date', how='left')
    .merge(visitorsBCummulative, left_on='date', right_on='date', how='left')
)

print(data.head(5))

ordersByUsersA = (
    orders[orders['group'] == 'A']
    .groupby('visitorId', as_index=False)
    .agg({'transactionId': pd.Series.nunique})
)
ordersByUsersA.columns = ['visitorId', 'orders']

ordersByUsersB = (
    orders[orders['group'] == 'B']
    .groupby('visitorId', as_index=False)
    .agg({'transactionId': pd.Series.nunique})
)
ordersByUsersB.columns = ['visitorId', 'orders'] 

pd.Series(0, index=np.arange(data['visitorsPerDateA'].sum() - len(ordersByUsersA['orders'])), name='orders')
 
[ordersByUsersA['orders'],pd.Series(0, index=np.arange(data['visitorsPerDateA'].sum() - len(ordersByUsersA['orders'])), name='orders')]

sampleA = pd.concat([ordersByUsersA['orders'],pd.Series(0, index=np.arange(data['visitorsPerDateA'].sum() - len(ordersByUsersA['orders'])), name='orders')],axis=0)

sampleB = pd.concat([ordersByUsersB['orders'],pd.Series(0, index=np.arange(data['visitorsPerDateB'].sum() - len(ordersByUsersB['orders'])), name='orders')],axis=0)

ordersByUsersA = (
    orders[orders['group'] == 'A']
    .groupby('visitorId', as_index=False)
    .agg({'transactionId': pd.Series.nunique})
)
ordersByUsersA.columns = ['visitorId', 'orders']

ordersByUsersB = (
    orders[orders['group'] == 'B']
    .groupby('visitorId', as_index=False)
    .agg({'transactionId': pd.Series.nunique})
)
ordersByUsersB.columns = ['visitorId', 'orders']

sampleA = pd.concat(
    [
        ordersByUsersA['orders'],
        pd.Series(
            0,
            index=np.arange(
                data['visitorsPerDateA'].sum() - len(ordersByUsersA['orders'])
            ),
            name='orders',
        ),
    ],
    axis=0,
)

sampleB = pd.concat(
    [
        ordersByUsersB['orders'],
        pd.Series(
            0,
            index=np.arange(
                data['visitorsPerDateB'].sum() - len(ordersByUsersB['orders'])
            ),
            name='orders',
        ),
    ],
    axis=0,
)

print("P-value:{0:.3f}".format(stats.mannwhitneyu(sampleA, sampleB)[1]))

print("Относительный прирост конверсии группы B по отношению к группе A:{0:.3f}".format(sampleB.mean() / sampleA.mean() - 1))