econuy 0.4.0

Download and process Uruguayan economic data.

ECON-UY

This project aims at simplifying gathering, processing and visualization (in the future) of Uruguay economic statistics. Data is retrieved from (mostly) government sources and can be transformed in several ways (converting to dollars, calculating rolling averages, resampling to other frequencies, etc.).

Installation

• PyPI:

pip install econuy

• Git:

git clone https://github.com/rxavier/uy-econ.git
python setup.py install

Usage

Read the documentation

The Session() class

Basics

This is the recommended entry point for the package. It allows setting up the common behavior for downloads, and holds the current working dataset.

from econuy.session import Session

session = Session(loc_dir="econuy-data", revise_rows="nodup", force_update=False)

The Session() object is initialized with the loc_dir, revise_rows, force_update and dataset attributes.

• loc_dir controls where data will be saved and where it will be looked for when updating. It defaults to "econuy-data", and will create the directory if it doesn't exist.
• revise_rows controls the updating mechanism. It can be an integer, denoting how many rows from the data held on disk to replace with new data, or a string. In the latter case, auto indicates that the amount of rows to be replaced will be determined from the inferred data frequency, while nodup replaces existing data with new data for each time period found in both.
• force_update controls whether whether to redownload data even if existing data in disk was modified recently.
• dataset holds the current working dataset and is initialized with an empty Pandas dataframe.

Methods

get() downloads the basic datasets. These are basically as provided by official sources, except various Pandas transformations are performed to render nice looking dataframes with appropiate column names, time indexes and properly defined values.

session.get(self, dataset: str, update: bool = True, save: bool = True, 
            override: Optional[str] = None, **kwargs)

Available options for the dataset argument are "cpi", "fiscal", "nxr", "naccounts", "labor", "rxr_custom", "rxr_official", "commodity_index", "reserves" and "fx_ops". Most are self explanatory but all are explained in the documentation.

override allows setting the CSV's filename to a different one than default (each dataset has a default, for example, "cpi.csv"). If you wanted CPI data:

df = session.get(dataset="cpi").dataset

Note that the previous code block accessed the dataset attribute in order to get a dataframe. Alternatively, one could also call the final() method after calling get().

get_tfm() gives access to predefined data pipelines that output frequently used data. These are based on the datasets provided by get(), but are transformed to render data that you might find more immediately useful.

session.get_tfm(self, dataset: str, update: bool = True, save: bool = True,
                override: Optional[str] = None, **kwargs)

For example, session.get_tfm(dataset="inflation") downloads CPI data, calculates annual inflation (pct change from a year ago), monthly inflation, and seasonally adjusted and trend monthly inflation.

Transformation methods take a Session() object with a valid dataset and allow performing preset transformation pipelines. For example:

df = session.get(dataset="nxr").decompose(flavor="trend", outlier=True, trading=False)

will return a the Session object, with the dataset attribute holding the trend component of nominal exchange rate.

Available transformation methods are

• resample() - resample data to a different frequency, taking into account whether data is of stock or flow type.
• chg_diff() - calculate percent changes or differences for same period last year, last period or at annual rate.
• decompose() - use X13-ARIMA to decompose series into trend and seasonally adjusted components.
• convert() - convert to US dollars, constant prices or percent of GDP.
• base_index() - set a period or window as 100, scale rest accordingly
• rolling() - calculate rolling windows, either average or sum.

X13 ARIMA binary

If you want to use the decompose() method you will need to supply the X13 binary (or place it somewhere reasonable and set x13_binary="search"). You can get it from here for Windows or from here for UNIX systems. For macOS you can compile it using the instructions found here (choose the non-html version) or use my version (working under macOS Catalina) from here.

Dataframe/CSV headers

Metadata for each dataset is held in Pandas MultiIndexes with the following:

1. Indicator name
2. Topic or area
3. Frequency
4. Unit/currency
5. Current or inflation adjusted
6. Base index period(s) (if applicable)
7. Seasonal adjustment
8. Type (stock or flow)
9. Cumulative periods

Word of warning

This project is heavily based on getting data from online sources that could change without notice, causing methods that download data to fail. While I try to stay on my toes and fix these quickly, it helps if you create an issue when you find one of these (or even submit a fix!).

What next

• I now realize this project would greatly benefit from OOP and plan to implement it next.
• Tests.
• CLI.
• Handling everything with column multi-indexes really doesn't seem like the best way to go around this.
• Automating data updates.