• What is A/B testing?
• Where can A/B testing be used?
• Key performance indicators (KPIs)
• meditation app
• Dataset 1: User demographics
• Dataset 2: User actions
• KPI: Conversion Rate
• Joining the demographic and subscription data

# Import pandas 
import pandas as pd

# Load the customer_data
customer_data = pd.read_csv("customer_data.csv")

# Load the app_purchases
app_purchases = pd.read_csv("inapp_purchases.csv")

# Print the columns of customer data
print(customer_data.columns)

# Print the columns of app_purchases
print(app_purchases.columns)

# Merge on the 'uid' field
uid_combined_data = app_purchases.merge(customer_data, on=['uid'], how='inner')

# Examine the results 
print(uid_combined_data.head())
print(len(uid_combined_data))

# Merge on the 'uid' and 'date' field
uid_date_combined_data = app_purchases.merge(customer_data, on=['uid', 'date'], how='inner')

# Examine the results 
print(uid_date_combined_data.head())
print(len(uid_date_combined_data))

• KPIs
• 1. Methods for calculating KPIs
• 2. Grouping Data: .groupby()
• 3. Aggregating data - mean price paid per group
• 4. Aggregate data: .agg()

# Calculate the mean purchase price 
purchase_price_mean = purchase_data.price.agg('mean')

# Examine the output 
print(purchase_price_mean)

# Calculate the mean and median purchase price 
purchase_price_summary = purchase_data.price.agg(['mean', 'median'])

# Examine the output 
print(purchase_price_summary)

# Calculate the mean and median of price and age
purchase_summary = purchase_data.agg({'price': ['mean', 'median'], 'age': ['mean', 'median']})

# Examine the output 
print(purchase_summary)

# Group the data 
grouped_purchase_data = purchase_data.groupby(by = ['device', 'gender'])

# Aggregate the data
purchase_summary = grouped_purchase_data.agg({'price': ['mean', 'median', 'std']})

# Examine the results
print(purchase_summary)

• Next Steps

# Compute max_purchase_date
max_purchase_date = current_date - timedelta(days=28)

# Filter to only include users who registered before our max date
purchase_data_filt = purchase_data[purchase_data.reg_date < max_purchase_date]

# Filter to contain only purchases within the first 28 days of registration
purchase_data_filt = purchase_data_filt[(purchase_data_filt.date <= 
                        purchase_data_filt.reg_date + timedelta(days=28))]

# Output the mean price paid per purchase
print(purchase_data_filt.price.mean())

# Set the max registration date to be one month before today
max_reg_date = current_date - timedelta(days=28)

# Find the month 1 values
month1 = np.where((purchase_data.reg_date < max_reg_date) &
                 (purchase_data.date < purchase_data.reg_date + timedelta(days=28)),
                  purchase_data.price, 
                  np.NaN)
                 
# Update the value in the DataFrame
purchase_data['month1'] = month1

# Group the data by gender and device 
purchase_data_upd = purchase_data.groupby(by=['gender', 'device'], as_index=False) 

# Aggregate the month1 and price data 
purchase_summary = purchase_data_upd.agg(
                        {'month1': ['mean', 'median'],
                        'price': ['mean', 'median']})

# Examine the results 
print(purchase_summary)