Predicting Visitor Purchases with BigQuery ML

GCP
PMLE
BigQuery

This Lab from Google PMLE course where I practice BigQuery ML

Author

Bruno Machado

Published

September 29, 2023

Predicting Visitor Purchases with BigQuery ML

Scenario:

Your data analyst team exported the Google Analytics logs for an ecommerce website into BigQuery and created a new table of all the raw commerce visitor session data for you to explore. Using this data, you’ll try to answer a few questions.

Question:

  1. Out of the total visitors who visited our website, what % made a purchase?
  2. What are the top 5 selling products?
  3. How many visitors bought on subsequent visits to the website?
  4. Looking at the initial data results, do you think time_on_site and bounces will be a good indicator of whether the user will return and purchase or not?

Exploring the data and answer the questions:

Question 1 : Out of the total visitors who visited our website, what % made a purchase?

To answer that question we gonig to dive total_purchasers / total_visitors :

R: 2.69%

#standardSQL

WITH visitors AS(
SELECT
COUNT(DISTINCT fullVisitorId) AS total_visitors
FROM `data-to-insights.ecommerce.web_analytics`
),
purchasers AS(
SELECT
COUNT(DISTINCT fullVisitorId) AS total_purchasers
FROM `data-to-insights.ecommerce.web_analytics`
WHERE totals.transactions IS NOT NULL
)
SELECT
  total_visitors,
  total_purchasers,
  total_purchasers / total_visitors AS conversion_rate
FROM visitors, purchasers

Question 1

Question 2 : What are the top 5 selling products?

To answer this question lets query and sum the productQuantity and return also the revenue


SELECT
  p.v2ProductName,
  p.v2ProductCategory,
  SUM(p.productQuantity) AS units_sold,
  ROUND(SUM(p.localProductRevenue/1000000),2) AS revenue
FROM `data-to-insights.ecommerce.web_analytics`,
UNNEST(hits) AS h,
UNNEST(h.product) AS p
GROUP BY 1, 2
ORDER BY revenue DESC
LIMIT 5;

Question 2

Question 3 : How many visitors bought on subsequent visits to the website?

This query will return two rows, one with total visitors that returned and buy and another one for not buy, the main logic in on that condition : IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit

# visitors who bought on a return visit could have bought on first as well

WITH all_visitor_stats AS (
SELECT
  fullvisitorid, # 741,721 unique visitors
  IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit
  FROM `data-to-insights.ecommerce.web_analytics`
  GROUP BY fullvisitorid
)
SELECT
  COUNT(DISTINCT fullvisitorid) AS total_visitors,
  will_buy_on_return_visit
FROM all_visitor_stats
GROUP BY will_buy_on_return_visit

Quesiton 3

Select features and create your training dataset

After explore the data lets create a dataset to create a Machine Learning model in BigQuery to predict whether or not a new user is likely to purchase in the future

Question:: Looking at the initial data results, do you think time_on_site and bounces will be a good indicator of whether the user will return and purchase or not?

In the lab it was provided that there are two important fields for that prediction

  • totals.bounces (whether the visitor left the website immediately)
  • totals.timeOnSite (how long the visitor was on our website)
SELECT
  * EXCEPT(fullVisitorId)
FROM
  # features
  (SELECT
    fullVisitorId,
    IFNULL(totals.bounces, 0) AS bounces,
    IFNULL(totals.timeOnSite, 0) AS time_on_site
  FROM
    `data-to-insights.ecommerce.web_analytics`
  WHERE
    totals.newVisits = 1)
  JOIN
  (SELECT
    fullvisitorid,
    IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit
  FROM
      `data-to-insights.ecommerce.web_analytics`
  GROUP BY fullvisitorid)
  USING (fullVisitorId)
ORDER BY time_on_site DESC
LIMIT 10;

Question 4

Select a BigQuery ML model type and specify options

  • before run we need to create a dataset ecommerce to save the model

Create the model

CREATE OR REPLACE MODEL `ecommerce.classification_model`
OPTIONS
(
model_type='logistic_reg',
labels = ['will_buy_on_return_visit']
)
AS
#standardSQL
SELECT
  * EXCEPT(fullVisitorId)
FROM
  # features
  (SELECT
    fullVisitorId,
    IFNULL(totals.bounces, 0) AS bounces,
    IFNULL(totals.timeOnSite, 0) AS time_on_site
  FROM
    `data-to-insights.ecommerce.web_analytics`
  WHERE
    totals.newVisits = 1
    AND date BETWEEN '20160801' AND '20170430') # train on first 9 months
  JOIN
  (SELECT
    fullvisitorid,
    IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit
  FROM
      `data-to-insights.ecommerce.web_analytics`
  GROUP BY fullvisitorid)
  USING (fullVisitorId)
;

Create the model

Evaluate classification model performance

In his classificatin problem we need :

  1. minimize the False Positive Rate (predict that the user will return and purchase and they don’t)
  2. maximize the True Positive Rate (predict that the user will return and purchase and they do)

In BigQuery ML, roc_auc is simply a queryable field when evaluating your trained ML model.

SELECT
  roc_auc,
  CASE
    WHEN roc_auc > .9 THEN 'good'
    WHEN roc_auc > .8 THEN 'fair'
    WHEN roc_auc > .7 THEN 'not great'
  ELSE 'poor' END AS model_quality
FROM
  ML.EVALUATE(MODEL ecommerce.classification_model,  (
SELECT
  * EXCEPT(fullVisitorId)
FROM
  # features
  (SELECT
    fullVisitorId,
    IFNULL(totals.bounces, 0) AS bounces,
    IFNULL(totals.timeOnSite, 0) AS time_on_site
  FROM
    `data-to-insights.ecommerce.web_analytics`
  WHERE
    totals.newVisits = 1
    AND date BETWEEN '20170501' AND '20170630') # eval on 2 months
  JOIN
  (SELECT
    fullvisitorid,
    IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit
  FROM
      `data-to-insights.ecommerce.web_analytics`
  GROUP BY fullvisitorid)
  USING (fullVisitorId)
));

Evaluate

Improve model performance with feature engineering

Add some new features and create a second machine learning model called classification_model_2:

  • How far the visitor got in the checkout process on their first visit
  • Where the visitor came from (traffic source: organic search, referring site etc.)
  • Device category (mobile, tablet, desktop)
  • Geographic information (country)

Create the NEW model:

CREATE OR REPLACE MODEL `ecommerce.classification_model_2`
OPTIONS
  (model_type='logistic_reg', labels = ['will_buy_on_return_visit']) AS
WITH all_visitor_stats AS (
SELECT
  fullvisitorid,
  IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit
  FROM `data-to-insights.ecommerce.web_analytics`
  GROUP BY fullvisitorid
)
# add in new features
SELECT * EXCEPT(unique_session_id) FROM (
  SELECT
      CONCAT(fullvisitorid, CAST(visitId AS STRING)) AS unique_session_id,
      # labels
      will_buy_on_return_visit,
      MAX(CAST(h.eCommerceAction.action_type AS INT64)) AS latest_ecommerce_progress,
      # behavior on the site
      IFNULL(totals.bounces, 0) AS bounces,
      IFNULL(totals.timeOnSite, 0) AS time_on_site,
      totals.pageviews,
      # where the visitor came from
      trafficSource.source,
      trafficSource.medium,
      channelGrouping,
      # mobile or desktop
      device.deviceCategory,
      # geographic
      IFNULL(geoNetwork.country, "") AS country
  FROM `data-to-insights.ecommerce.web_analytics`,
     UNNEST(hits) AS h
    JOIN all_visitor_stats USING(fullvisitorid)
  WHERE 1=1
    # only predict for new visits
    AND totals.newVisits = 1
    AND date BETWEEN '20160801' AND '20170430' # train 9 months
  GROUP BY
  unique_session_id,
  will_buy_on_return_visit,
  bounces,
  time_on_site,
  totals.pageviews,
  trafficSource.source,
  trafficSource.medium,
  channelGrouping,
  device.deviceCategory,
  country
);

New Model

Evaluate



#standardSQL
SELECT
  roc_auc,
  CASE
    WHEN roc_auc > .9 THEN 'good'
    WHEN roc_auc > .8 THEN 'fair'
    WHEN roc_auc > .7 THEN 'not great'
  ELSE 'poor' END AS model_quality
FROM
  ML.EVALUATE(MODEL ecommerce.classification_model_2,  (
WITH all_visitor_stats AS (
SELECT
  fullvisitorid,
  IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit
  FROM `data-to-insights.ecommerce.web_analytics`
  GROUP BY fullvisitorid
)
# add in new features
SELECT * EXCEPT(unique_session_id) FROM (
  SELECT
      CONCAT(fullvisitorid, CAST(visitId AS STRING)) AS unique_session_id,
      # labels
      will_buy_on_return_visit,
      MAX(CAST(h.eCommerceAction.action_type AS INT64)) AS latest_ecommerce_progress,
      # behavior on the site
      IFNULL(totals.bounces, 0) AS bounces,
      IFNULL(totals.timeOnSite, 0) AS time_on_site,
      totals.pageviews,
      # where the visitor came from
      trafficSource.source,
      trafficSource.medium,
      channelGrouping,
      # mobile or desktop
      device.deviceCategory,
      # geographic
      IFNULL(geoNetwork.country, "") AS country
  FROM `data-to-insights.ecommerce.web_analytics`,
     UNNEST(hits) AS h
    JOIN all_visitor_stats USING(fullvisitorid)
  WHERE 1=1
    # only predict for new visits
    AND totals.newVisits = 1
    AND date BETWEEN '20170501' AND '20170630' # eval 2 months
  GROUP BY
  unique_session_id,
  will_buy_on_return_visit,
  bounces,
  time_on_site,
  totals.pageviews,
  trafficSource.source,
  trafficSource.medium,
  channelGrouping,
  device.deviceCategory,
  country
)
));

Evaluate new model

Predict which new visitors will come back and purchase

Lets predict the probability that a first-time visitor will make a purchase in a later visit

SELECT
*
FROM
  ml.PREDICT(MODEL `ecommerce.classification_model_2`,
   (
WITH all_visitor_stats AS (
SELECT
  fullvisitorid,
  IF(COUNTIF(totals.transactions > 0 AND totals.newVisits IS NULL) > 0, 1, 0) AS will_buy_on_return_visit
  FROM `data-to-insights.ecommerce.web_analytics`
  GROUP BY fullvisitorid
)
  SELECT
      CONCAT(fullvisitorid, '-',CAST(visitId AS STRING)) AS unique_session_id,
      # labels
      will_buy_on_return_visit,
      MAX(CAST(h.eCommerceAction.action_type AS INT64)) AS latest_ecommerce_progress,
      # behavior on the site
      IFNULL(totals.bounces, 0) AS bounces,
      IFNULL(totals.timeOnSite, 0) AS time_on_site,
      totals.pageviews,
      # where the visitor came from
      trafficSource.source,
      trafficSource.medium,
      channelGrouping,
      # mobile or desktop
      device.deviceCategory,
      # geographic
      IFNULL(geoNetwork.country, "") AS country
  FROM `data-to-insights.ecommerce.web_analytics`,
     UNNEST(hits) AS h
    JOIN all_visitor_stats USING(fullvisitorid)
  WHERE
    # only predict for new visits
    totals.newVisits = 1
    AND date BETWEEN '20170701' AND '20170801' # test 1 month
  GROUP BY
  unique_session_id,
  will_buy_on_return_visit,
  bounces,
  time_on_site,
  totals.pageviews,
  trafficSource.source,
  trafficSource.medium,
  channelGrouping,
  device.deviceCategory,
  country
)
)
ORDER BY
  predicted_will_buy_on_return_visit DESC;

Predict

More options to create model

Using Deep learning and neural network :

Using AutoMl and XGBoost:

AutoML

auto_ml model

XGBoost

XGBoost model