morganpare/ecx_analytics

Project to develop analytics suite for ethiopian commodities exchange, specifically the coffee commodity # Ethiopian Coffee Exchange (ECX) Analytics Suite This project was designed to give a basic set of analytics tools for historical data relating to the Ethiopian Coffee Exchange. The initial scope is to provide a proof-of-concept market price prediction API alongside a set of visualisations tool for understanding market characteristics in the past 7 years. ## Installation This repo is python based, so you should use whichever environment manager you prefer to install the relevant packages found in the `requirements.txt` file. You should also download the relevant dataset from Kaggle Datasets here kaggle.com . Either run the `make_data_directory.sh` file or create a data directory in the home directory with the following structure: - data - bronze - silver - gold - feature_store Extract the excel files into the bronze directory and delete the original `.zip` file. ## Usage The project has three core functionalities: * Data Processor * Price Predictor * Visualisation ### Data Processor To run the data processing step, use the `data_processor_main.py` script found within the respective layer of the `ecx_analytics` package. This will populate the data directory with the appropriate data. This data model mimics the delta lake pattern databricks.com in a lightweight manner. ### Price Predictor The price prediction functionality uses a swagger-based endpoint, this can be running by issuing the command `python -m swagger_server` whilst in the home directory. This will spawn a local server which can be accessed via the browser. There is also a Dockerfile which can be used to spawn a server. The application path is `{server}/ecx-analytics` . The swagger UI can be found at `{server}/ecx-analytics/ui` Once the server is running, you can issue REST API calls to the server endpoints. Specifying the coffee symbol and target d …