Scribd
Upload
39 views

Assignment: Machine Learning Engineer: Problem Description 1 (NLP)

The document outlines two machine learning problems for a machine learning engineer to solve. The first problem involves building a sentiment analysis model for user reviews of medicines that can classify reviews as positive, negative, or neutral and generate sentiment scores, incorporating information from complex reviews mentioning multiple medicines. The second problem involves building an image classification model to categorize images into four classes: food, miscellaneous, attire, and decoration/signage, using a training dataset of 5,000 images and evaluating on a test set of 983 images. The engineer has 48 hours to solve one or both problems by putting all code in a GitHub repo and submitting the trained model and output.

Uploaded by

himanshu aditya
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
39 views

Assignment: Machine Learning Engineer: Problem Description 1 (NLP)

The document outlines two machine learning problems for a machine learning engineer to solve. The first problem involves building a sentiment analysis model for user reviews of medicines that can classify reviews as positive, negative, or neutral and generate sentiment scores, incorporating information from complex reviews mentioning multiple medicines. The second problem involves building an image classification model to categorize images into four classes: food, miscellaneous, attire, and decoration/signage, using a training dataset of 5,000 images and evaluating on a test set of 983 images. The engineer has 48 hours to solve one or both problems by putting all code in a GitHub repo and submitting the trained model and output.

Uploaded by

himanshu aditya
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Assignment: Machine Learning Engineer 

Problem Description 1 (NLP) 

You have been selected to work on the dataset of a pharmaceutical company. They have scrapped the 
dataset of user reviews about various medicines.  
Each review is associated with a medicine name. But the reviews provided by the users are complex in 
nature and some reviews contain information about multiple medicines as well.  
For example: 
​Medicine A review​: “​It works well on fever. Medicine A is better than Medicine B”​ . 
The overall review is positive for Medicine A, but it is negative for Medicine B.  
Your task is to build an Sentiment Analysis model. It should have following feature:  
● For a given medicine, your model should classify review as Positive, Negative or Neutral. 
● It should generate sentiment scores based on overall data i.e. incorporating the complex reviews 
information. 
 
Dataset Link: https://drive.google.com/drive/folders/1kPzur0zPvspZa5zcohGCBb8UCLrsvvwk?usp=sharing 
 
Problem Description 2 (Computer Vision) 

Build a object classification system for 4 categories: Food, Misc, Attire, Decorationandsignage 
 
● We have provided a “train.csv” file with two columns, one having an image name and the other has 
a label associated with each image. The size of training dataset is 5000 images 
● There is a test folder with 983 images in it and an associated csv file with Image name. 
● You are supposed to build an image classification model using Train Images only.  
● Do the evaluation on test images and submit the output in a csv file (same format as train.csv) with 
the name “submission.csv” 
 
Don't try to use any Test image during the training phase. Also don’t manually modify the results in the 
“submission.csv” file. If we notice anything suspicious, your assignment will be rejected. 
 
Dataset Link: https://drive.google.com/drive/folders/1iX-1r-Oj8Zfj9wZfm8mEjZJbrN-LbN0O?usp=sharing 
 
Submission details: 

● You are supposed to solve the problems in a 48 hour time duration. Clock starts from the time you 
receive this email (tik-tok-tik-tok) 
● We are expecting you to solve one problem statement (as per your expertise). Brownie points if 
you solve both the problems in the given time duration. ;) 
● Put all the code in a Github public repo and share the repo URL in reply to the assignment email. 
● Make a proper Readme.md file with a detailed description of your approach. 
● Submit both your trained model and the output. 
● Choose the evaluation metric accordingly 
 
All the best.  
 
 
 

You might also like