MachineHack, in association with Ugam, a Merkle company, is back again with an exciting hiring hackathon — ‘uHack Sentiments 2.0: Decode Code Words’ — for the data science and machine learning community from Dec 22, 2021, to Jan 10, 2022. Participate in this exclusive challenge and stand a chance to get hired by an analytics and technology services company company & win exciting cash prizes.
Ugam, a Merkle company, is a leading analytics and technology services company whose customer-centric approach delivers impactful business results for large corporations by leveraging data, technology, and expertise. The company consistently delivers superior, impactful results through the right blend of human intelligence and AI.
With 3300+ people spread across locations worldwide, Ugam has successfully deployed its services to create success stories across industries like retail and consumer brands, high tech, BFSI, distribution, and market research and consulting. Over the past 21 years, Ugam has been recognised by several firms, including Forrester and Gartner, named the No.1 data science company in India by Analytics Insight, and certified as a Great Place to Work®.
The company had shared its plans to hire over 1,300 analytics and technology professionals in India. In the past seven months, Ugam has added 1,200 plus people including analysts, DevOps, software engineers, technical architects, and technical leads with experience in Adobe, AWS, and Google Cloud platforms.
So, this is the chance for data scientists, ML practitioners, analytics professionals, and enthusiasts to showcase their skills and expertise and build a meaningful career in the data analytics industry.
The challenge starts on December 22, 2021 at 6:00 PM.
Click here to participate in the hackathon.
Problem Statement & Description
The last two decades have witnessed a significant change in how consumers purchase products and express their experience/opinions in reviews, posts, and content across platforms. These online reviews are not only useful to reflect customers’ sentiment towards a product but also help businesses fix gaps and find potential opportunities which could further influence future purchases.
To that end, Ugam, a Merkle company, in association with MachineHack, is inviting the entire machine learning community to develop a machine learning model that can analyse customers’ sentiments based on their reviews and feedback.
Unlike typical sentiment analysis problems, this challenge isn’t as simple as it appears to be and thus gives an opportunity for the participants to come up with some innovative solutions and gauge their data science knowledge. This hackathon is not your average sentiment analysis problem, rather goes beyond.
Participants of this hackathon will get to analyse and deep dive into the natural language text (reviews) and bucket them based on their discussion topics. Plus, they will get to analyse the overall sentiment of the business to make tangible decisions.
MachineHack and Ugam have created a training dataset of 6136 rows with 14 columns and a testing dataset of 2631 rows with 14 columns to solve the sentiment analysis problem. The dataset will be a mix of customer reviews for products across categories and retailers.
The participants need to develop a model that can bucket the future reviews in their respective topics, where a review can talk about multiple topics and overall polarity, i.e. positive/negative sentiment.
The hackathon demands a few prerequisite skills such as text pre-processing – Lemmatization, tokenisation, N-Grams and other relevant methods, along with multi-class classification and optimising Log Loss.
Submission Guidelines
The participants must submit a .csv/.xlsx file with exactly 2631 rows with 12 columns . The submission will return an ‘Invalid Score’ if any extra columns or rows are presented.
Sklearn models support the predict_proba() method to generate the predicted values.
Evaluation Criteria
The evaluation of the hackathon will be done using the Log Loss metric. One can use sklearn.metric.log_loss to calculate the same.
The hackathon will also support private and public leaderboards, where the public leaderboard will be evaluated on 70% of test data. On the other hand, the private leaderboard will be available at the hackathon’s end and assessed on 100% of test data.
The final score will represent the score achieved based on the ‘Best Score’ on the public leaderboard.
Prizes
Ugam will select the top three (3) winners based on the given criteria. The prize money is for interested candidates willing to get interviewed/hired by Ugam.
First Prize: INR 40,000
Second Price: INR 20,000
Third Prize: INR 10,000
The hackathon will end on January 10, 2021 at 6:00 PM.
Click here to participate in the hackathon.
Dataset Details
- Train.csv – 6136 rows x 14 columns
- Test.csv – 2631 rows x 14 columns
- Sample Submission.csv — Please check the ‘Evaluation’ section on MachineHack Page for more details on generating a valid submission.
What to predict?
- Topics (components, delivery and customer support, design and aesthetics, dimensions, features, functionality, installation, material, price, quality, and usability)
- Polarity (positive/negative)
Evaluation criteria: Log Loss metric
Note: The target variables are all encoded in the training dataset for convenience. Please submit the test results in a similar encoded fashion for us to evaluate your results.
Attribute Description:
- ID
- Review
- Components
- Delivery and Customer Support
- Design and Aesthetics
- Dimensions
- Features
- Functionality
- Installation
- Material
- Price
- Quality
- Usability
- Polarity
Skills:
- Text Pre-processing – Lemmatization, Tokenization, N-Grams and other relevant methods
- Multi-Class and Multi Label Classification
- Optimizing Log Loss
Note:
- The prize money will be for the interested candidates who are willing to get interviewed or hired by Ugam.
- Winner are requested to come to the Machine Leaning Developers Summit 2022, happening at Bangalore, for receiving the prize money.
