Adding predictive variables

Intermediate Predictive Analytics in Python

Nele Verbiest

Senior Data Scientist @PythonPredictions

Predictive variables

Demographics:
- Age
- Gender
- Living place
Spending behaviour
Watching behaviour
Product usage
Surfing behaviour
Payment information

Timeline compliant predictive variables (1)

Timeline compliant predictive variables (2)

Adding lifetime

# Reference date
reference_date = datetime.date(2018,4,1)

# Add lifetime to the basetable
basetable["lifetime"] = reference_date - basetable["member_since"]
print(basetable.head())

donor_id member_since lifetime
1        2015-02-03   1153
2        2016-01-30   729
3        2016-02-23   768

Adding preferred contact channel (1)

donor_id start_valid_date end_valid_date  contact_channel
1        2014-02-03       2016-03-04      "phone"
1        2016-03-04       2016-05-08      "e-mail"
2        2016-02-23       2026-02-23      "e-mail"

reference_date = datetime.date(2018,4,1)

# Select lines compliant with reference data
contact_channel_reference_date 
    = living_places[
      (contact_channel["start_valid_date"]<=reference_date) &                                             
      (living_places["end_valid_date"]>reference_date)]

Adding preferred contact channel (2)

# Add contact channel place to the basetable
basetable = 
    pd.merge(
     basetable, 
     living_places_reference_date[["donor_ID","contact_channel"]],
     on="donor_ID"
    )
print(basetable.head())

donor_id contact_channel
1        "phone"
2        "phone"
3        "e-mail"

Let's practice!

Intermediate Predictive Analytics in Python