Using data from the Nairobi Coffee Exchange for coffee prices from 2014-2018, scrape data from PDFs, and forecast future coffee prices for the 2018-2019 session.
Uses Tabula for data scraping and Prophet for time-series forecasting.
For easy viewing:
-
Create and activate a new virtual environment:
$ conda create -n coffee_env python=3$ conda activate coffee_env
-
Install project dependencies:
$ conda install gcc; conda install -c conda-forge fbprophet$ pip install -r requirements.txt
- PDF_to_CSV.ipynb: Jupyter notebook that parses and cleans the data from the PDF files into a CSV file.
- Predicting_Coffee_Prices.ipynb: Jupyter notebook that plots the monthly average prices for grade AA, AB, and C coffee, and forecasts the coffee prices for 2018-2019.
- Projecting_Coffee_Prices_in_Kenya.pdf: PDF report summarizing findings.
- monthly_coffee_prices.csv: Resulting dataset from parsing PDF files.
- pdfs/: Directory of original PDF files.
- predictions/: Directory of CSV files containing predictions for the 2018-2019 session for grade AA, AB, and C coffee.