Go by Bus is an application for storing/analyzing communication data from Warsaw's open data platform.
It uses current GPS location of trams, bus stops locations & timetables to achieve following goals:

1. Store current & historical tram positions for analysis and ML [done]
2. Store timetables data for analysing line delays [done]
3. Visualize current & historical locations of queried tram using Google Maps API [done]
4. Current tram locations as a stream of data from Kafka [done]
5. Finding anomalies in traffic and calculate communication delays in Spark [TODO 1]
6. Stores nearest & historical weather info thanks to yr.no API and enriches delay analysis [TODO 2]
7. Visualizes timetables for selected line [TODO 3]

• We are using Microservices with Java 8 + Spring Cloud based on Docker
• CQRS architecture is applied (heart of system is Apache Kafka)
• Apache Kafka for real-time locations stream
• Configuration is stored in central Spring Config Service
• Service logs + GC logs are connected to ELK (ElasticSearch + LogStash + Kibana). But no visualisations yet.
• Data storing done in MongoDB
• Docker as a container service, and docker-compose for getting up the environment for now.
• Apache Spark as a main data analysis tool - module SparkPositionAnalyzer need a lot of development thought
• Simple long-time-running master version is deployed to AWS using docker-machine
• Gradle as a build tool

1. Refactor and develop more completed Spark queries
2. Introduce new datasource - Weather data from yr.no
3. Create separate service for timetables data based on GraphQL
4. Introduce cross-service user tracking with Zipkin
5. Prepare Kibana log visualizations
6. Introduce more complex orchestrating tool ei. Kubernetes
7. Introduce node monitoring - Zabbix

1. Install Docker and docker-compose
2. Increase vm.max_map_count for your machine due to ELK requirements
3. Create your account and generate API key on Warsaw's open data platform.
4. Put your API key in secret-keys.properties file in main dir as a WARSAW_API_KEY= property
5. docker compose up -d in main dir
6. Have fun :)

Bare in mind that solution is pretty complexed and drains a lot of resources. On i7 + 16 GB RAM it's ok. Some clustering will be introduced in future for sure