Script 709: SBA Auto Pause & Re enable Budget via Dimension Tags

Purpose

The Python script automates the pausing and resuming of advertising campaigns based on monthly budget constraints and dimension tags, monitoring spending bi-hourly.

To Elaborate

The script is designed to manage advertising campaigns by automatically pausing them when the month-to-date (MTD) spending reaches a predefined monthly budget, which is stored in dimension tags. It also resumes campaigns when the spending falls below the budget threshold. The script operates on a bi-hourly basis, ensuring that campaigns are paused or resumed promptly according to the budget constraints. A safety margin is applied to account for potential delays and non-linear spending patterns. The script processes data locally or on a server, using a pickled data source for local execution. It calculates the MTD spending for each budget group and compares it against the monthly budget, adjusting campaign statuses accordingly.

Walking Through the Code

Configuration and Setup
- The script begins by defining a configurable parameter, BUDGET_CAP_SAFETY_MARGIN, which sets a safety margin for budget calculations.
- It checks whether the script is running locally or on a server, loading necessary data from a pickle file if running locally.
Data Preparation
- The script loads the primary data source into a DataFrame and ensures data types are correct, converting budget columns to numeric and date columns to date types.
- It fills any missing values with blanks to prevent comparison errors.
Budget Calculation and Campaign Status Recommendation
- The script calculates the MTD spending for each budget group and identifies campaigns that exceed their budget, recommending them for pausing.
- It also identifies campaigns that can be resumed if their spending is below the budget threshold and they have a recorded pause date.
Campaign Status Update
- The script updates the campaign status to “Paused” or “Active” based on the recommendations, ensuring that only campaigns marked for traffic are affected.
- It records the pause date for campaigns that are paused and clears it for those that are resumed.
Output and Debugging
- The script identifies changes in campaign status and prepares the output DataFrame with updated statuses.
- If running locally, it writes the output and debug information to CSV files for further analysis.

Vitals

Script ID : 709
Client ID / Customer ID: 309909744 / 14196
Action Type: Bulk Upload
Item Changed: Campaign
Output Columns: Account, Campaign, Status, SBA Pause Date, SBA Recommended Status
Linked Datasource: M1 Report
Reference Datasource: None
Owner: Michael Huang (mhuang@marinsoftware.com)
Created by Michael Huang on 2024-02-21 21:58
Last Updated by ascott@marinsoftware.com on 2024-02-23 21:26

> See it in Action

Python Code

##
## name: Intraday Budget Cap via Dimensions
## description:
##  Pause campaigns when MTD spend reaches Monthly Budget (stored in Dimensions)
## 
## author: Michael S. Huang
## created: 2024-02-09
## 

##### Configurable Param #####
# Define how close MTD spend can get to Monthly Budget before being Paused
#  - compensates for lag in system
#  - compendates for non-linearity in intraday spend
BUDGET_CAP_SAFETY_MARGIN = 0.02 # set to 2%
##############################


########### START - Local Mode Config ###########
# Step 1: Uncomment download_preview_input flag and run Preview successfully with the Datasources you want
download_preview_input=False
# Step 2: In MarinOne, go to Scripts -> Preview -> Logs, download 'dataSourceDict' pickle file, and update pickle_path below
pickle_path = '/Users/mhuang/Downloads/pickle/generic_budget_cap_intraday_datasource_dict.pkl'
# Step 3: Copy this script into local IDE with Python virtual env loaded with pandas and numpy.
# Step 4: Run locally with below code to init dataSourceDict

# determine if code is running on server or locally
def is_executing_on_server():
    try:
        # Attempt to access a known restricted builtin
        dict_items = dataSourceDict.items()
        return True
    except NameError:
        # NameError: dataSourceDict object is missing (indicating not on server)
        return False

local_dev = False

if is_executing_on_server():
    print("Code is executing on server. Skip init.")
elif len(pickle_path) > 3:
    print("Code is NOT executing on server. Doing init.")
    local_dev = True
    # load dataSourceDict via pickled file
    import pickle
    dataSourceDict = pickle.load(open(pickle_path, 'rb'))

    # print shape and first 5 rows for each entry in dataSourceDict
    for key, value in dataSourceDict.items():
        print(f"Shape of dataSourceDict[{key}]: {value.shape}")
        # print(f"First 5 rows of dataSourceDict[{key}]:\n{value.head(5)}")

    # set outputDf same as inputDf
    inputDf = dataSourceDict["1"]
    outputDf = inputDf.copy()

    # setup timezone
    import datetime
    # Chicago Timezone is GMT-5. Adjust as needed.
    CLIENT_TIMEZONE = datetime.timezone(datetime.timedelta(hours=-5))

    # import pandas
    import pandas as pd
    import numpy as np

    # Printing out the version of Python, Pandas and Numpy
    # import sys
    # python_version = sys.version
    # pandas_version = pd.__version__
    # numpy_version = np.__version__

    # print(f"python version: {python_version}")
    # print(f"pandas version: {pandas_version}")
    # print(f"numpy version: {numpy_version}")

    # other imports
    import re
    import urllib

    # import Marin util functions
    from marin_scripts_utils import tableize, select_changed
else:
    print("Running locally but no pickle path defined. dataSourceDict not loaded.")
    exit(1)
########### END - Local Mode Setup ###########

today = datetime.datetime.now(CLIENT_TIMEZONE).date()

# primary data source and columns
inputDf = dataSourceDict["1"]
RPT_COL_ACCOUNT = 'Account'
RPT_COL_CAMPAIGN = 'Campaign'
RPT_COL_CAMPAIGN_STATUS = 'Campaign Status'
RPT_COL_PUB_COST = 'Pub. Cost $'
RPT_COL_SBA_TRAFFIC = 'SBA Traffic'
RPT_COL_SBA_BUDGET_GROUP = 'SBA Strategy'
RPT_COL_SBA_GROUP_MONTHLY_BUDGET = 'SBA Campaign Budget'
RPT_COL_SBA_PAUSE_DATE = 'SBA Pause Date'
RPT_COL_SBA_RECOMMENDED_STATUS = 'SBA Recommended Status'

# output columns and initial values
BULK_COL_ACCOUNT = 'Account'
BULK_COL_CAMPAIGN = 'Campaign'
BULK_COL_STATUS = 'Status'
BULK_COL_SBA_PAUSE_DATE = 'SBA Pause Date'
BULK_COL_SBA_RECOMMENDED_STATUS = 'SBA Recommended Status'

# Make inputDf a copy of original to keep dataSourceDict.pkl pristine
originalDf = dataSourceDict["1"]
inputDf = originalDf.copy()

# define some intermediate columns
COL_MTD_BUDGET_GROUP_SPEND = 'mtd_budget_group_spend'

# define Status values
VAL_STATUS_ACTIVE = 'Active'
VAL_STATUS_PAUSED = 'Paused'
VAL_BLANK = ''

print("inputDf shape", inputDf.shape)
print("inputDf info", inputDf.info())

## force expected types
# Convert RPT_COL_SBA_MONTHLY_BUDGET to numeric, coercing errors to NaN
inputDf[RPT_COL_SBA_GROUP_MONTHLY_BUDGET] = pd.to_numeric(inputDf[RPT_COL_SBA_GROUP_MONTHLY_BUDGET], errors='coerce')
# Replace NaN values with 0.0 if that's the desired behavior
inputDf[RPT_COL_SBA_GROUP_MONTHLY_BUDGET].fillna(0.0, inplace=True)
# Force RPT_COL_SBA_PAUSE_DATE to be Date type
inputDf[RPT_COL_SBA_PAUSE_DATE] = pd.to_datetime(inputDf[RPT_COL_SBA_PAUSE_DATE], errors='coerce').dt.date

# HACK: replace nan with empty strings so comparison doesn't fail
inputDf.fillna(VAL_BLANK, inplace=True)





# Calculate MTD Budget Group Spend
inputDf[COL_MTD_BUDGET_GROUP_SPEND] = inputDf.groupby(RPT_COL_SBA_BUDGET_GROUP)[RPT_COL_PUB_COST].transform('sum')

# Recommend to Pause camapigns with MTD Budget Group Spend over Monthly Budget (by a margin)
has_monthly_group_budget = inputDf[RPT_COL_SBA_GROUP_MONTHLY_BUDGET] > 0.0
over_spent_campaigns = inputDf[COL_MTD_BUDGET_GROUP_SPEND] >= inputDf[RPT_COL_SBA_GROUP_MONTHLY_BUDGET] * (1 - BUDGET_CAP_SAFETY_MARGIN)
campaigns_to_pause = has_monthly_group_budget & over_spent_campaigns

inputDf.loc[campaigns_to_pause, 'pause'] = 1
print(f"campaigns_to_pause count: {sum(campaigns_to_pause)}")
if campaigns_to_pause.any():
    print("campaigns_to_pause campaigns", tableize(inputDf.loc[campaigns_to_pause].head()))

inputDf.loc[ campaigns_to_pause, \
             RPT_COL_SBA_RECOMMENDED_STATUS \
           ] = VAL_STATUS_PAUSED
 
# Recommend to reactivate campaigns with MTD Budget Group Spend under Monthly Group Budget (by a margin)
# but limited to campaigns with SBA Pause Date populated 10 digit date
under_spent_campaigns = inputDf[COL_MTD_BUDGET_GROUP_SPEND] < inputDf[RPT_COL_SBA_GROUP_MONTHLY_BUDGET] * (1 - BUDGET_CAP_SAFETY_MARGIN)
sba_paused_campaigns = inputDf[RPT_COL_SBA_PAUSE_DATE].astype('str').str.len() >= 10
campaigns_to_resume = under_spent_campaigns & sba_paused_campaigns

inputDf.loc[campaigns_to_resume, 'resume'] = 1
print(f"campaigns_to_resume count: {sum(campaigns_to_resume)}")
if campaigns_to_resume.any():
    print("campaigns_to_resume", tableize(inputDf.loc[campaigns_to_resume].head()))

inputDf.loc[ campaigns_to_resume, \
             RPT_COL_SBA_RECOMMENDED_STATUS  \
           ] = VAL_STATUS_ACTIVE

## Actually taffic PAUSE

should_traffic = inputDf[RPT_COL_SBA_TRAFFIC].astype(str).str.lower() == 'traffic'

should_traffic_pause = should_traffic & \
                       (inputDf[RPT_COL_SBA_RECOMMENDED_STATUS] == VAL_STATUS_PAUSED) & \
                       (inputDf[RPT_COL_SBA_RECOMMENDED_STATUS] != inputDf[RPT_COL_CAMPAIGN_STATUS])


inputDf.loc[should_traffic_pause, 'traffic_pause'] = 1
print(f"should_traffic_pause count: {sum(should_traffic_pause)}")
if should_traffic_pause.any():
    print("should_traffic_pause campaigns", tableize(inputDf.loc[should_traffic_pause].head()))


inputDf.loc[should_traffic_pause, RPT_COL_CAMPAIGN_STATUS] = inputDf.loc[should_traffic_pause, RPT_COL_SBA_RECOMMENDED_STATUS]
inputDf.loc[should_traffic_pause, RPT_COL_SBA_PAUSE_DATE] = today.strftime('%Y-%m-%d')

## Actually taffic RESUME

should_traffic_resume = should_traffic & \
                       (inputDf[RPT_COL_SBA_RECOMMENDED_STATUS] == VAL_STATUS_ACTIVE) & \
                       (inputDf[RPT_COL_SBA_RECOMMENDED_STATUS] != inputDf[RPT_COL_CAMPAIGN_STATUS]) & \
                       sba_paused_campaigns

inputDf.loc[should_traffic_resume, 'traffic_resume'] = 1
print(f"should_traffic_resume count: {sum(should_traffic_resume)}")
if should_traffic_resume.any():
    print("should_traffic_resume campaigns", tableize(inputDf.loc[should_traffic_resume].head()))

inputDf.loc[should_traffic_resume, RPT_COL_CAMPAIGN_STATUS] = inputDf.loc[should_traffic_resume, RPT_COL_SBA_RECOMMENDED_STATUS]
inputDf.loc[should_traffic_resume, RPT_COL_SBA_PAUSE_DATE] = VAL_BLANK

print(f"select_changed with inputDf shape {inputDf.shape} and originalDf shape {originalDf.shape}")


# only include changed rows in bulk file
(outputDf, debugDf) = select_changed(inputDf, \
                                    originalDf, \
                                    diff_cols = [ \
                                        RPT_COL_CAMPAIGN_STATUS, \
                                        RPT_COL_SBA_RECOMMENDED_STATUS, \
                                    ], \
                                    select_cols = [ \
                                        RPT_COL_ACCOUNT, \
                                        RPT_COL_CAMPAIGN, \
                                        RPT_COL_CAMPAIGN_STATUS, \
                                        RPT_COL_SBA_RECOMMENDED_STATUS, \
                                        RPT_COL_SBA_PAUSE_DATE, \
                                    ], \
                                    merged_cols=[RPT_COL_ACCOUNT, RPT_COL_CAMPAIGN] \
                                    )


changed = (debugDf[RPT_COL_CAMPAIGN_STATUS+'_new'] != debugDf[RPT_COL_CAMPAIGN_STATUS+'_orig']) | \
          (debugDf[RPT_COL_SBA_RECOMMENDED_STATUS+'_new'] != debugDf[RPT_COL_SBA_RECOMMENDED_STATUS+'_orig'])

debugDf.loc[changed, 'changed'] = 1
print(f"changed count: {sum(changed)}")
if changed.any():
    print("changed campaigns", tableize(debugDf.loc[changed].head()))

# remember to use Bulk column header for Status
outputDf = outputDf.rename(columns = { \
                RPT_COL_CAMPAIGN_STATUS: BULK_COL_STATUS \
                })

print("outputDf shape", outputDf.shape)
print("outputDf", tableize(outputDf.tail(5)))

## local debug
if local_dev:
    output_filename = 'outputDf.csv'
    outputDf.to_csv(output_filename, index=False)
    print(f"Local Dev: Output written to: {output_filename}")

    debug_filename = 'debugDf.csv'
    debugDf.to_csv(debug_filename, index=False)
    print(f"Local Dev: Debug written to: {debug_filename}")

Post generated on 2024-11-27 06:58:46 GMT

21 Feb 2024

« Script 707: Campaign Anomaly Detection USIM University of the Southwest Script 711: Ad Group CPA Performance Outlier »

MarinOne Scripts Creator's Corner