Python script

  • Posted:
  • Proposals: 1
  • Remote
  • #693388
  • Awarded
Darien I.
has already sent a proposal.
  • 0


Experience Level: Intermediate
Please take a few minutes to read the task in details.

the task requires in implementation of algorithm of Perceptron using Python programming language, ((without)) using any built-in (classification algorithm) from library, but, using numpy and scipy libraries is allowed for the data accessing such as (scipy.sparse) or (numbly.array).

Please download the attached. On that file, you will find four files: train.positive , train.negative , test.positive , and test.negative .
These files correspond to the positive and negative train/test reviews will be using in this task. Each line in each file represents a review using a set of features. We will be using both unigram and bigram (concatenated using two underscores) features to represent a review. A review is represented using a bag-of-features. Moreover, each feature is counted only once, giving a boolean valued feature representation (i.e. a set of features for each review).

The Task?

1. We need to write a program to load the train/test instances (positive/negative) from the train/test files.
2. Implement a binary Perceptron classifier and measure the classification accuracy on the test instances. Classification accuracy is defined as the percentage of the total number of correctly classified instances to the total number of test instances.
3. Plot the train error rate and test error rate against the number of iterations. According to your plot, what would be the ideal number of iterations to terminate the training?

* please, check the (algorithm.png) attached, to check the perceptron algorithm required to be used.

Kind regards

New Proposal

Create an account now and send a proposal now to get this job.

Sign up

Clarification Board Ask a Question

    There are no clarification messages.