Authors : Cyril Nérin - Hugo Rialan - Alexandre Perbet
The objective of the project is to predict markets stocks using online regression tools The stock exchange rates used are as follows : Google, Facebook, Amazon, Total, Gazprom, Alibaba, BNP Paribas, Ferrari
We first developed a batch regression solution. Then, we developed an online solution using river and Kafka.
The project was carried out on Jupyter notebooks to improve the visibility of the results obtained.
3 notebooks are available :
-
The first one contains the data ingestion and an implementation of the ARIMA algorithm for the batch part.
-
The second one contains the online learning prediction algorithm implemented with River.
-
The third one allows to display and plot the obtained results.
It is assumed that Zookeeper is running default on localhost:2181 and Kafka on localhost:9092 before running the notebooks.