Star Coders Compete (PREDICTATHON)

Date/Time
Date(s) - 07/10/2021 - 12/10/2021
10:00 am - 5:00 pm

Categories No Categories


Registration Link

https://forms.office.com/r/EiGxYnTzWh

Problem Statement:

Given certain input parameters regarding Student personal, academic and placement details, predict the placement outcome of any given student at the end of 6th semester.

  • Programming Language to be used: only Python 3.8
  • Maximum size of zip file allowed: 50MB
  • Maximum permitted execution time: 10secs.
  • We will be executing your code on 8GB RAM machine – so please ensure that the code submitted will execute within these constraints for us to get the output.

Recommended packages

  • The python environment that will be used for evaluation is follows:
    • h5py==2.10.0
    • jupyter==1.0.0
    • Keras==2.4.3
    • Keras-Preprocessing==1.1.2
    • numpy==1.19.5
    • pandas==1.2.4
    • scikit-learn==0.24.1
    • scipy==1.6.2
    • seaborn==0.11.1
    • tensorflow==2.4.1
    • torch==1.8.1
  • Please note that we will not support the use of any other packages (or versions) that are not present in this file. You may use different packages for your training phase. However, kindly ensure that your test code during submission does not depend on packages other than the mentioned above.
  • In case of errors such as packages not found, the submission will be disqualified.
  • It is also to be noted that we will not be using GPUs during our tests. Therefore, please refrain from using packages that depend on GPU computing (NOTE: tensorflow is supported; tensorflow-gpu is not).

Input Data:

We are providing the link to the Dataset which contains historic data of student studied in the past. This dataset may be used by the candidate teams to train an ML model or come up with a data analytics based algorithm that can perform the required prediction as mentioned above

Dataset download:

CSV Files for Student details: Click here to download

Data Description (quoted directly from Cricsheet page: https://cricsheet.org/downloads/):

The CSV File provided follows the “NEW” format.
The “new” format consists of a single row for each student studied in our college.

The first row of each CSV file contains the headers for the file, with each subsequent row providing details on a single delivery. The headers in the file are:

General 

Student Id
Date of Birth
SSLC Mark
SSLC School Studied
SSLC Medium
Diploma Percentage
Diploma Studied College
Diploma Specialization
HSC Mark
HSC Medium
HSC Group
SWS/MQ/FN
Address (Pincode Only)
Father Occupation
Mother Occupation
Community
Religion
Nationality
First Graduate (Y/N)
Cutoff
Gender
Physically Challenged (Y/N)

 

Academic 

GPA (1st Semester to 8th Semester)
Number of Arrears

Placement 

Student Placed In (Company Name)
CTC

Test Case Input:

The following will be provided as input test case data:

The candidates may consume the data and preprocess or convert them in any manner for making it work as per their developed model.

Input format

student_id
1001

Evaluation:

Points scored by each team = R2 error value (sum of square of error) between the actual score and predicted score (The lower the points, better is the prediction outcome).

Submission:

File structure

SubmissionFormat.zip contains two python source files main.py and predictor.py.
(The lower the points, better is the prediction outcome).
predictor.py: In the sample format, predictor.py contains a definition predictPlacement ().
main.py: This is a test file. It is recommended to NOT edit this file. Once you implement the predictor.py, you can test the functionality of your implementation.

How to implement:

You can customize the implementation of predictPlacement () in predictor.py. A list of recommended python based ML packages can be found below. If you are making a model, then create your training script separately and train it against the training data given on the contest page. At the end of the training, save your model as an appropriate file. In predictPlacement, load your model and do the necessary computation by feeding the input. You can customize your implementation by adding more classes and definitions to support predictPlacement. The expected return value is an integer Let model.xyz be the saved ML model, then save it as a sibling document to predictor.py. Whenever the model is loaded, use the relative path to read the file.

Leave a Reply

Your email address will not be published. Required fields are marked *