A Beginner’s Guide to DBT (data build tool) — Part 2 : Setup guide and tips

Installation and Data pre-processing

This is my second episode. If you are new and want to know what are the key power of DBT, you can visit my first episode on this #dbtdataseries.

Getting started with data build tool (dbt) on Google BigQuery

1) Set up DBT

Homebrew

What is Homebrew? Homebrew is an open-source software package manager that makes it easier to install software on macOS. It’s like AppStore for installation.

1.1 Install Homebrew. Then, run:

brew update
brew install git
brew tap fishtown-analytics/dbt

1.2 Test your installation with dbt --version.

Take note: The most time consuming part is brew update, and sometimes when you runbrew update it seems that it just stopped there, but it is actually running behind the scenes. So please do not cancel brew update or close your laptop; otherwise it might take longer!

2) Create a BigQuery project

2.1 Go to the BigQuery Console

2.2 Create a new project

2.3 Head back to the the BigQuery Console, and ensure your new project is selected. Please take note of your project location because later on you will need to put it into your dbt profiles.ymlconfig file.

3) Generate BigQuery credentials

3.1 Go to the BigQuery credential wizard. Ensure to choose your correct project name.

For example: shiba-ecommerce is my project name

3.2 Click Create Credentials > Help me choose. Generate credentials with the following options:

  • Which API are you using? BigQuery API
  • What data will you be accessing? Application data
  • Are you planning to use this API with App Engine or Compute Engine? No, I’m not using them
  • Service account details — Service account name: dbt-user
  • Grant this service account access to the project — Role: BigQuery Admin
  • Grant users access to this service account — Service account admin roles: I put my gmail.

Click Done.

3.3 Go back to the BigQuery credential wizard.

3.4 Under the Service Accounts, click your newly-created service account > Go to Keys> Click Show Domain-wide Delegation> Click Create new key> Choose JSON file > Download the JSON file and save it in an easy-to-remember spot, with a clear filename (example: dbt-user-creds.json)

4) Create data set

  • Once you select your project name, click View actions > Create data set > Name your data set as accordingly.
Create my `shiba-ecommerce` data set

Take note: Please take note of your project location because later on you will need to put it into your dbt profiles.yml config file. For example, I chose Singapore (asia-southeast1) as my data location.

5) Clean Data

  • As mentioned above, the dataset I took is Brazilian E-Commerce Public Dataset by Olist. For this project, I will use all 9 .csv files except for olist_geolocation_dataset.csv
  • When you upload data to BigQuery from .csv files, the maximum file size is 10 MB. So what I did is to trim a couple of rows so that all files are under 10 MB before the upload.
  • In this tutorial, I will demo the upload of 4 files — olist_customers_dataset.csv olist_order_items_dataset.csv olist_order_reviews.csv product_category_translation.csv because each of which has a different type of error and workaround.

5.1 — olist_customers_dataset.csv

  • This file is the easiest. In your newly created dataset, click Create table > Choose Upload and select the olist_customers_dataset.csv > Name your table as customers> Tick Auto-detect schema .
NOTE: For all .csv files, I change all `olist` to `shiba`. It’s just my project naming convention.

The final result should look like this.

`customers` Table has been successfully imported (table in preview mode)

5.2 — olist_order_items_dataset.csv

Let’s move on to the second file. Let’s select olist_order_items_dataset.csv and follow the steps as the first file.

  • Error: You may likely encounter this error — seeing null values in the table order_items when importing to BigQuery. However, let’s try to click the last page, you will see non-null values.
  • Reason: The null values are caused by the empty rows in the .csv file.
order_items table preview has error: NULL values
  • Workaround: Hence, based on my experience, the best practice is to open .csv files in your code editor to check if there are any empty rows, and delete all empty rows. In the image below, I dragged the olist_order_items_dataset.csv in Visual Code Studio and delete all of the empty rows from line 51237 onwards.
Drag my `olist_order_items_dataset.csv` to Visual Code Studio
  • Error: You may also encounter the problem related to the TIMESTAMP data type. Supposedly the correct date time format in BigQuery is yyyy/mm/dd hh:mm:ss . However, as it can be seen from the column shipping_limit_date , the date time format is 00dd/mm/yy hh:mm:ss 0018-01-18 02:48:00 UTC . This is not compatible in BigQuery.
order_items table preview has error: wrong date time format
  • Reason: In .csv all of the date time columns, for example, the shipping_limit_date appears as dd/mm/yyyy hh:mm:ss . Hence, we will need to change the date format to yyyy/mm/dd hh:mm:ss to be compatible with that of BigQuery.
  • Workaround: To change the date format, select all rows from the shipping_limit_date column (shift + command + down arrow for Mac) (shift + ctrl + down arrow for Windows) > Click More numbers format > Under Type , change from d/m/yy hh:mm to yyyy/mm/dd hh:mm:ss (please see my image below for reference).
In excel, change the date format of the shipping_limit_date column

Now let’s try to import again. The result table should look like this.

`order_items` Table has been successfully cleaned and imported (table in preview mode)

5.3 — Upload olist_order_reviews.csv

Let’s move on to the third file — Now I am going to create order_reviews table from olist_order_reviews.csv file.

  • Error: Then I got the error below.
Error while reading data, error message: Error detected while parsing row starting at position: 1765. Error: Missing close double quote (") character.Error while reading data, error message: CSV processing encountered too many errors, giving up. Rows: 12; errors: 1; max bad: 0; error percent: 0
The Error Message
  • Reason: It is due to special characters such as the double quote character" in the review_comment_message column.
  • Workaround: Therefore when you import this file, you will need to tick Allow quoted newlines under Advanced options.
Tick Allow quoted newlines

And here is the final result.

`order_reviews` Table has been successfully imported (table in preview mode)

5.4 — product_category_name_translation.csv

Yayyy here’s the last file I would like to demo!

  • Error: When I create a product_category_name_translation table from product_category_name_translation.csv file, I get incorrect column names such as string_field_x like shown in the image below.
Table Schema of product_category_translation table
  • Reason: I did some research, and found the workaround from Samet Karadag (thank you!)
  • Workaround: We will add a dummy integer column int in the product_category_name_translation table. Then let’s try to create the product_category_name_translation table again. Now you will see that column names are recognised correctly.
  • The next step is to remove that int column by creating a new table using this query.
CREATE table `<project_id>.<dataset>.<table>` as
SELECT * except(int)
FROM `<project_id>.<dataset>.<table_ext>`

In details
project id:
shiba-ecommerce
dataset:
shiba_dataset
table_ext: product_category_name_translation (the initial table)
table:
product_category_name_trans (the new table after removing int column)

For example, here is my query

Tadaa your final result is here!

`product_category_name_trans` Table has been successfully cleaned and imported (table in preview mode)

Final Word

In summary, I have covered the installation part (how to set up DBT, create a BigQuery project at the BigQuery Console, generate BigQuery credentials at BigQuery credential wizard, create a data set) and the data pre-processing (clean data and upload data to BigQuery).

In my next episode, I will demonstrate dbt run and dbt test in details together with my struggles and workaround as well!

Stay tuned!

Data-driven, strategic professional with a passion for driving user acquisition and product performance. Eager to make a social impact in this VUCA world.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store