People and Acquisition Facts

About

This application takes constituent data, email data, and subscription data to generate two files: a "people" file containing a subset of data on each constituent from the three source files, and an "acquisition facts" file that aggregates statistics on when people in the dataset were acquired.

Setup

This application requires Python 3 and Pandas. Please see instructions for Pandas installation. The recommended installation method for Pandas is to install as part of the Anaconda distribution, which also installs Python. Ensure python has been added to your PATH environment variable, or replace every call to python below with the full path to python, e.g. C:\Users\alexa\anaconda3\python.exe.

Quick Start

To get started right away, open a command prompt in the directory containing generate_people.py, then run python generate_people.py --help for usage. For more detailed instructions, see below.

Run

• This application is run via command prompt.

• To run from the directory which contains generate_people.py, run: python generate_people.py.

  • To run from a directory other than the one which contains generate_people.py, provide the path to the application, e.g. python path\to\generate_people.py.

• When the application is run with no additional parameters, it will search for constituent data, email data, and subscription data in the current directory, as cons.csv, cons_email.csv, and cons_email_chapter_subscription.csv respectively.

  • To specify a different input source for constituent data, use the -c flag followed by a full filepath (including filename), e.g. -c path\to\some_cons.csv
  • To specify a different input source for email data, use the -e flag followed by a full filepath (including filename), e.g. -c path\to\some_cons_email.csv
  • To specify a different input source for subscription data, use the -s flag followed by a full filepath (including filename), e.g. -c path\to\some_cons_email_chapter_subscription.csv

• When the application is run with no additional parameters, it will generate output files people.csv and acquisition_facts.csv in the current directory.

  • To specify a different output destination for people data, use the -p flag followed by a full filepath (including filename), e.g. -p path\to\some_people.csv
  • To specify a different output destination for acquisition facts data, use the -a flag followed by a full filepath (including filename), e.g. -a path\to\some_acquisition_facts.csv
  • Any directories specified as part of the path to output data must exist already.

• If the output files exist already, they will be overwritten.

Usage Examples

These are usage examples only. Absolute file paths will differ on your machine.

Run from a directory containing generate_people.py and the input files with their default names (generate_people.py, cons.csv, cons_email.csv, and cons_email_chapter_subscription.csv). Output to default files (people.csv and acquisition_facts.csv in the current directory):

python generate_people.py

Run from a directory containing input files with their default names but not generate_people.py. Output to default files.

python C:\workspace\people\generate_people.py

Run from a directory containing generate_people.py and the input files with their default names. Output to the default file location, but rename people data to my_people.csv.

python generate_people.py -p my_people.csv

Run from a directory containing generate_people.py and a sub-directory input containing cons.csv, cons_email.csv, and cons_email_chapter_subscription.csv. Output people data and acquisition facts data to a sub-directory output. (Note that the output directory must already exist.):

python generate_people.py -c input\cons.csv -e input\cons_email.csv -s input\cons_email_chapter_subscription.csv -p output\people.csv -a output\acquisition_facts.csv

Project Assumptions

The following assumption were made to complete this exercise. Under normal circumstances, these points would be clarified with colleagues or with the client:

• Constituents must be captured in the people data even if they have no associated primary email address.
  • If this assumption is incorrect, the merge of the constituents data with the emails data should use how='inner' instead of how='left'.
• The people.csv schema should be exactly as described, without key values.
  • In a real implementation, it might be useful to preserve cons_id or cons_email_id, and possibly to generate a new key, people_id.
• The acquisition_facts file should be a .csv file with a header line.
  • The exercise did not specify the file format or whether it should have a header line.
• Any order for acquisition_facts is acceptable.
  • Current implementation sorts by descending number of acquisitions. It may be useful to sort chronologically by date.
• Expected structure of input data is known to the user of this tool.
  • If this assumption is incorrect, additional validation should be performed on the input files, and more details added to the readme.