Aaron Plocharczyk's project update 1 post

by Aaron Plocharczyk

14 Jun 2017

Here’s the current trinket:



Progress since last class:
I stayed up all night workig since last class because I have to fly to and be in a wedding this weekend. I am on track and working ahead to make time for other things I need to do soon. I have created review, file, trainingfile, and model classes. Training file extends file, and files are made up of reviews. I also made and inperpreter module to allow the user to type their own commands, to interact with the program in any way at any time. The user can type “help” to learn about these commands after they select a training file. The training file I am using right now is called “imdb-train.txt”. Additionally, I made a printer module to print well formatted tables and other forms of output easily. I still want to find time to let the user test their model and a testing file. I hope to do this before next class. I also want to let the user visualize their data sometimes.

OLD Milestones:
By 6/14:

  • program asks user for a training file and creates a “trainingFile” object from it

    By 6/15:
  • program counts total number of occurrences for each term
  • program calculates correlation of terms with a sentiment

    By 6/19:
  • program prints and visualizes correlation and total number of occurrences for each term
  • program allows user to set a minimum threshold for total number of occurrences for each term
  • program allows user to delete particular terms

    By 6/22:
  • program asks user for a testing file and creates a “testingFile” object from it
  • program tests current model on the test file and saves the results locally to refer to in the next iteration of the program
  • program prints a visual representation of the test results as compared to previous tests
  • program allows user to type “help” at any time to learn about what they should do
  • program repeats

    Stretch Goals:
  • program allows user to save model in a comma separated format
  • program allows user to open saved models
  • program allows user to alter saved models
  • program allows user to test with saved models




CURRENT Milestones:
By 6/14:

  • program asks user for a training file and creates a “trainingFile” object from it
  • program can print two dimensional list structures as well formatted tables

    By 6/15:
  • program counts total number of occurrences for each term
  • program counts total number of occurrences for a specified term
  • program calculates correlation of terms with a sentiment
  • program calculates correlation of a specified term with a sentiment
  • program allows user to perform any action with a the training set at any time by letting them type their own commands

    By 6/19:
  • program prints and visualizes correlation and total number of occurrences for each term
  • program allows user to set a minimum threshold for total number of occurrences for each term
  • program allows user to delete particular terms
  • program allows user to restore particular terms that have been deleted

    By 6/22:
  • program asks user for a testing file and creates a “testingFile” object from it
  • program tests current model on the test file and saves the results locally to refer to in the next iteration of the program
  • program prints a visual representation of the test results as compared to previous tests
  • program allows user to type “help” at any time to learn about what they should do
  • program allows user to get further details about the commands listed in the help menu
  • program repeats

    Stretch Goals:
  • program allows user to save model in a comma separated format
  • program allows user to open saved models
  • program allows user to alter saved models
  • program allows user to test with saved models

    This is definitely an amitious project. I have spent 13 hours on it so far and still have a lot to do. I hope I can do enough with the time that I have to produce something that feels like a finished product.
I am a senior at UNC Chapel Hill taking my last couple courses this summer. Find Aaron Plocharczyk on Twitter, Github, and on the web.