AirBNB Use Case

If you do not have a DataPlatform account, log in at https://eu.dataplatform.ovh.net/ with your email and password

Data Source Configuration

Datacatalog

The goal is to obtain two CSV files from the S3 endpoint: reviews.csv and listings.csv. Go to the Datacatalog section and click on “Create a source”.

Step 2: Create a New S3‑Compatible Storage Source

Enter the following information to create a new S3 storage source.

Field	Value
Access key	f15278c3829a4a03b1c4ed0ee779410c
Secret key	d4cfeb306fe44cd6b8cf78d8f9932fe1
Endpoint	s3.gra.io.cloud.ovh.net
Bucket	rbnbfortechlab
Region	gra

Step 3: Analyze the Data with the Analyzer

Go to the Analyzer section and run the analysis on both files.
See the impact of the blueprints.

• Edit the attribute review_per_month of the blueprint and replace the replace by rule with the value 0 when the field is empty.

Lakehouse Manager Configuration

Step 1: Create Tables from Data Sources

Create two tables from the two previous data sources: listings and reviews

Step 2: Create Working Tables

Create a third empty table with 9 attributes, named aggregation

Attribute	Type	Category
id	Double	Measure
latitude	String	Measure
longitude	String	Measure
room_type	String	Measure
price	Double	Measure
date	Date	Dimension
comment	String	Measure
emotion	String	Measure
avantage	String	Measure

You can now build the table. (Build at the top right)

Duplicate this table and name it review3031122023

Rebuild all tables.

Creating Actions and Workflows: Data Processing Engine

You already have two predefined load actions: they were created when the tables were created from the data sources.

Step 1: Create an Aggregation Action

• Create an aggregation action and join on id and listing_id, putting the aggregation into the aggregation table.
• listings.id = reviews.listing_id
• Map the correct fields and remove emotion and avantage from the aggregation. Execute the action.

After the action runs successfully, you can verify in Lakehouse manager ⇒ explorer that your data is there.

Step 2: Custom Action

Now we only want the comments from December 30 and 31 2023 so we can later run sentiment analysis via an AI endpoint.

• Build a custom action in Python with pandas and put all comments from these two dates into the table review3031122023.

Link to the stub code

https://rbnbfortechlab.s3.gra.io.cloud.ovh.net/workonlytwodays.py.

Link to the solution

https://rbnbfortechlab.s3.gra.io.cloud.ovh.net/onlytwodays.py.

Run the action.

Creating a Delete Action

Create a delete action that will empty all tables.

Creating a Workflow

We need four steps:

• Empty all tables
• Load the raw data
• Aggregate the first two tables
• Run the custom action

Verify that the workflow works correctly by executing it. Then in explorer verify that you have exactly 2 847 rows in the table.