The app's main goal is to parse the most frequent key rate per month in a convinient DataFrame structure.
Bank of Russia provides SOAP 1.1-API interface, so the "zeep" library was an ideal choice to parse data. The data gets processed and cached into Panda's DataFrame.
"fromDate, ToDate YYYY-MM-DD" - Two main variables needed to set the desired date range. ToDate variable must accept the last day of the month since it also be used in ToDates POST XML parameter, which is inclusive.
"dates" - Creates a range of dates. freq parameter sets the frequency. The default "MS" alias defines month start frequency.
"df_sales" - Holds the revenue data to showcase a simple use case.
"sales_rates_2024" - Contains parsed rates and dummy revenue values.
get_cbr_key_rate(fromDate:str, ToDate:str) -> pd.DataFrame
The function parses key rates and caches obtained data with SQLite. The data gets passed into the DataFrame through read_xml method where an unnecessary hours part gets truncated.
df.groupby(df.index.to_period('M')).agg(lambda x: x.mode().iloc[0]) - given some month, this line filters the most frequently met key rate and stores it for that month.
key_rate_and_sales(func:"a function to get the key rates") -> pd.DataFrame
Unites the key rates from previous function with the given sales from df_sales.
There are two graphs to visualize the data. First bar-graph depicts the revenue and key side by side and the second graph are two splitted line graphs that depicts revenue and key rates on their own.
The parsed data is used for statistical hypothesis testing to check if there are any relations between company's revenue and the key rate.
Python 3.13.3. zeep, pandas, and matplotlib as the main modules. requirements.txt holds all the necessary dependencies.
The code was written within Jupyter Notebook on Windows 11 Home, version 24H2, 26100.4061 OS build.