List Uniquification¶

Project Goals¶

This engineering effort invites you to investigate how to use various discrete structures, such as the dictionary and the set, to extract the unique values contained inside of a list of strings. In addition to implementing and testing the functions that perform list uniquification, you will produce the majority of the functions for the program's command-line interface and input processing. You will also implement functions that calculate the reduction in size and the percent reduction in size for a list of strings with an unknown amount of redundancy. To experimentally assess the efficiency of the datauniquifier program that you implement, this project also invites you to conduct an experiment to study, for instance, which uniquification process is best at reducing a list's size.

Project Access¶

If you are a student enrolled in a Computer Science class at Allegheny College, you can access this assignment by clicking the link provided to you in Discord. Once you click this link it will create a GitHub repository that you can clone to your computer by following the general-purpose instructions in the description of the technical skills. Specifically, you will need to use the git clone command to download the project from GitHub to your computer. Now you are ready to add source code and documentation to the project!

Expected Output¶

To get started on this engineering effort, please make sure that you read the blog post entitled Fastest Way to Uniquify a List in Python. As part of this assignment, you are going to implement at least three ways to remove the duplicate values from one of the columns in a large input file that contains automatically generated data. Here is a sample of the data that you input into the program that you will implement as part of this assignment:

dana74@mahoney-perez.com,"Administrator, charities/voluntary organisations"
nathanjohnson@davila.net,Software engineer
pbush@gmail.com,"Journalist, newspaper"
timothy75@chang.com,Osteopath
gsparks@yahoo.com,"Psychologist, clinical"
daniel39@gmail.com,Logistics and distribution manager
jason85@ward.com,Logistics and distribution manager
jacobwalton@hotmail.com,Television camera operator
markmcgee@hernandez-roberts.com,IT sales professional
shannon35@allen.com,Ecologist

When the program accepts this type of input it will also accept a specific column for which it should eliminate duplicates through the process of uniquification. For instance, when the program is run with the command poetry run datauniquifier --approach listcomprehension --column 1 --data-file input/data.txt then it will remove all of the duplicates from column 1 in the data file that stores the job descriptions of specific individuals. Alternatively, if the program was run with the command poetry run datauniquifier --approach listcomprehension --column 0 --data-file input/data.txt then it will remove all of the duplicates from column 0 in the data file that stores the email addresses of specific individuals. If we run the first of these two previous commands then the program will produce output like this:

The chosen approach to uniquify the file is: listcomprehension

The data file that contains the input is: input/data.txt

The data file contains 50000 data values in it!

🚀 Let's do some uniquification!

🔍 So, was this an efficient approach to uniquifying the data?

The function 'unique_listcomprehension' took: 0.0063 sec

Estimated overall memory according to the operating system:
   37.921875 megabytes

🔍 So, did this remove a lot of duplicate data?

The number of values removed from the data: 1155
The percent reduction due to uniquification: 2.31%

Adding Functionality¶

One of your tasks for this project is to address all of the TODO markers in the analyze, extract, and main modules of the datauniqifier program. After you have completed all of the TODO markers inside of the provided Python source code, you should execute the program in a variety of configurations so as to determine the influence that the size of the input data set, the procedure chosen for performing uniquification, and the type of data that is input into the uniquification procedure has on the memory and time efficiency of the process and the amount of reduction achieved by a specific configuration.

To automatically generate data sets of different sizes, you can use the CSV Faker tool that relies on the Faker Package package with a command like csvfaker --rows 50000 email job > data.txt. Note that this command will create a data file called data.txt that contains two columns, the first for an email and the second for a job. It is also important to note that this command will generate a data set that contains a total of 50,000 individual records of data. Please bear in mind that running the csvfaker program in this fashion may, depending on the performance characteristics of your laptop, require a long time to run. Using the aforementioned approach for running the csvfaker program you should generate different data files and then use them as the input to the datauniquifier program. While everyone learning to be a proactive programmer is encouraged to use the csvfaker tool to generate their own data sets, you can complete this project's required tasks by using the provided data.txt file in the input directory.

Along with varying the size of the data, your experiments should also consider how the removal of redundant data values varies depending on the type of the data input into your tool. You can do this by running the program with both --column 0 and --column 1. Finally, you should notice that the uniquify.py file contains a total of three different procedures for performing uniquification, with more approaches outlined in the blog post called Fastest Way to Uniquify a List in Python. You should make sure to run the datauniquifier with at least the three required ways to remove duplication, checking to see if different approaches vary in terms of their memory consumption and execution time. Along with noticing the trends in the data sets that you collect, you should also aim to explain why these trends are evident, leveraging your knowledge of how the Python programming languages uses discrete structures such as the set.

The evaluation metrics for the efficiency of the datauniquifier program are as follows: (i) execution time of the approach, (ii) estimated memory overhead of the entire Python program, (iii) reduction in the size of the column of data, and (iv) percent reduction in the size of the column of data. As you are working to understand each of these evaluation metrics, make sure that you review the following Python functions that respectively calculate the reduction and percent reduction in the size of the data. In the calculate_reduction function, line 3 calculates and returns the difference in size between the length of the list_final and the list_start, with larger values suggesting that there was a greater reduction in the size of the list. Finally, lines 3 through 5 in calculate_percent_reduction use the output of the calculate_reduction function to compute the percent reduction when the size of list_start is compared to that of list_final, with higher values indicating a greater overall reduction in size.

1
2
3

def calculate_reduction(list_start, list_final):
    """Calculate the reduction in the size of the list."""
    return len(list_start) - len(list_final)

1
2
3
4
5

def calculate_percent_reduction(list_start, list_final):
    """Calculate the percent reduction in the size of the list."""
    reduction = calculate_reduction(list_start, list_final)
    percent_reduction = (reduction / len(list_start)) * 100
    return percent_reduction

As you conduct the experiment using the datauniqifier program, you should ask yourself what type of values would suggest that the tool worked well. For instance, if you want to reduce the overall size of a data set through the process of uniquification, is it better to have a large or a small value for these two evaluation metrics calculated by calculate_reduction and calculate_percent_reduction? Remember, part of your goal for this assignment is to evaluate how the different configurations of the datauniquifier program influence these four evaluation metrics! To accomplish this task you will need to run the datauniqifier with different command-line arguments and record the time and memory overhead data that it reports in your terminal window.

One noteworthy aspect of this program is that it uses the getattr function to "construct" an executable version of a Python function when provided with the name of the function, as described in this StackOverflow discussion. After reading the discussion on StackOverflow, make sure that you understand the source code line unique_result_list = function_to_call(data_column_text_list). You should also notice that, instead of accepting as input the full name of a function, this program accepts the name of the approach and then builds up the name of the function. Can you find and understand the source code that completes this task? Finally, this approach adopts a different approach to recording the execution time of the three functions that perform uniquification, leveraging the timing "decorator" described in the following function. Make sure that you review the following StackOverflow discussion to understand how this approach works!

def timing(function):
    """Define a profiling function for execution time."""
    @wraps(function)
    def wrap(*args, **kw):
        ts = time()
        result = function(*args, **kw)
        te = time()
        print("The function %r took: %2.4f sec" % (function.__name__, te - ts))
        return result

    return wrap

Running Checks¶

As you continue to add and confirm the correctness of datauniquifier's functionality, you should study the source code in the pyproject.toml file. This file contains the specification of several tasks that will help you to easily run checks on your Python source code. Now, you can run commands like poetry run task lint to automatically run all of the linters designed to check the Python source code in your program and its test suite. You can also use the command poetry run task black to confirm that your source code adheres to the industry-standard format defined by the black tool. If it does not adhere to the standard then you can run the command poetry run fixformat and it will automatically reformat the source code. By following a tutorial, you can configure VS Code to use the black tool to automatically reformat the source code when you save a file.

Along with running tasks like poetry run task lint, you can leverage the relevant instructions in the technical skills to run the command gatorgrade --config config/gatorgrade.yml to check your work. If your work meets the baseline requirements and adheres to the best practices that proactive programmers adopt you will see that all the checks pass when you run gatorgrade. You can study the config/gatorgrade.yml file in your repository to learn how the GatorGrade program runs GatorGrader to automatically check your program and technical writing. If your program has all of the anticipated functionality, you can run the command poetry run task test-silent and see that the test suite produces output like the following. It is important to note that datauniqifier comes with three test suites, each of which, as shown below, should pass so as to establish a confidence in program correctness.

tests/test_analyze.py ......                                         [ 54%]
tests/test_extract.py ..                                             [ 72%]
tests/test_uniquify.py ...                                           [100%]

Project Reflection¶

Once you have finished both of the previous technical tasks, you can use a text editor to answer all of the questions in the writing/reflection.md file. For instance, you should provide the output of the Python program in a fenced code block, explain the meaning of the Python source code segments that you implemented, and answer all of the other questions about your experiences in completing this project. The finished version of your reflection should fully describe the influence that the size of the input data set, the procedure chosen for performing uniquification, and the type of data that is input into the uniquification procedure has on the memory and time efficiency of the process and the amount of size reduction achieved by a specific configuration.

Project Assessment¶

Since this project is an engineering effort, it is aligned with the evaluating and creating levels of Bloom's taxonomy. You can learn more about how a proactive programming expert will assess your work by examining the assessment strategy. From the start to the end of this project you may make an unlimited number of reattempts at submitting source code and technical writing that meet every aspect of the project's specification.

Seeking Assistance¶

Emerging proactive programmers who have questions about this project are invited to ask them in either the GitHub discussions forum or the Proactive Programmers Discord server. Before you ask your question, please read the advice concerning how to best participate in the Proactive Programmers community. If you find a mistake in this project, please describe it and propose a solution by creating an issue in the GitHub Issue Tracker.