Exploring the Kaggle dataset on Forex rates in the past 20 years
Data (Foreign_Exchange_Rates.csv) obtained from https://www.kaggle.com/brunotly/foreign-exchange-rates-per-dollar-20002019
Blog post at https://medium.com/@prannerta100/a-foray-into-forex-what-do-the-past-20-years-tell-us-dfd80e63249f
Dependencies:
- Pandas (for data manipulations)
- Seaborn (for plotting)
- matplotlib (for plotting)
- sklearn (for scaling, principal components analysis)
Forex rates are important indicators of the economic health of a country. Political events, etc. also play a major role in determining the forex rate of a country. On the flip side, multi-national corporations might base their investment decisions on forex rates.
The data is in .csv format, and has the Forex rates (per USD) of various countries (including Euro zone) over the past 20 years (2000-2019) as its columns.
Data wrangling:
- Renaming columns to shorter, more accessible names: for example, 'Euro Area - Euro/US$' gets shortened to 'Euro_Area'.
- Exchange rates are converted to numeric values, so that we can perform arithmetic manipulations with them.
- Dates are converted into Pandas standard Date Time format
- Certain rows of exchange rate values in the dataset are labelled as "ND". Given that we have enough valid numerical data already, our conclusions won't change significantly if we drop rows containing such non-numerical values.
Data analysis: Pearson's correlation coeeficient and principal components analysis (PCA) are the 2 methods used for analyzing the data.
Downloads: Look at the Jupyter notebook in this folder. In order to be able to run the Jupyter notebook code, make sure you also download the Foreign_Exchange_Rates.csv data file in this repository. Feel free to provide feedback!