Statistics of real estate data in HCMC

Processing data of 200k houses in HCMC, raw data after scraping from nhatot.vn is put into HDFS, then transform with Spark and load into redshift data warehouse. BI tool for data using Superset.

Technology

visualization: Superset backend: Python, Hadoop, Spark, S3, Redshift.

Usage

1. Crawler module.

A class written to crawl houses in the last 2 months on nhatot.vn. Using headers and random request times.

from crawler.module import Crawler

bot = Crawler()
data = bot.run()

Result:

2. Put raw data to Data Lake Hadoop HDFS.

Turn on HDFS

ssh localhost
start-all.sh

Put local files to HDFS

hdfs dfs -mkdir /lake
hdfs dfs -copyFromLocal /home/jayson/raw310323.json /lake/

3. Transform with Spark and Load to S3.

Combine multi json files in hdfs.

python3 spark/combine.py

Filter data and load to s3.

python3 spark/process.py

4. Put data to Redshift data warehouse.

Run query in file redshift/run.sql

5. Connect with superset.

6. (Alternative) Mongo Visualization.

Because of the limitation in deploying superset to production, it takes a lot of resources, so you can refer to the data visualized on mongo atlas.

Demo (mongo)

https://house.vietdoo.engineer/