AI for the People, by the People

Just like you don’t need to be an engineer to drive a car, it’s our firm belief that you don’t need to be a programmer or a data scientist to use AI. As such, Ikigai organized its first ever case competition on the Ikigai product, which is an end-to-end AI-powered platform for data operations, together with Sardar Vallabhbhai National Institute of Technology [SVNIT, Surathttps://www.svnit.ac.in],an institute of national importance in India.  

The event brought together a total of 228 teams consisting 465 students form diverse branches of sciences and engineering. Noteworthy, the vast majority of the participants - over 80% was from non-computer science background and got exposed to the idea of using AI for the first time. The competition provided a platform for students to showcase their skills in data analytics, and to propose solutions to real-world problems.

‍

The Challenge

How much do the homes in Bengaluru, the capital city of Karnataka, cost? A solution to this question, a real and complex housing prices prediction problem, was what the students were challenged to deliver.

Bangaluru is known as the "Silicon Valley of India" due to its strong IT industry and start-up ecosystem. The multitude of factors such as city’s millennial population, dynamic culture, wonderful environment, and abundance of employment opportunities make it a challenging yet fascinating task of estimating the cost of a particular home.  

In the real-world scenario, this problem is typically solved by either relying on the expert intuition of an experienced real estate broker who would be able to estimate the ballpark, or determined through the appraisal process, again, by the industry professionals. Finally, given the right data, a data scientist can of course solve this problem  

However, equipped with data and the right predictive techniques – or no-code tools like Ikigai – virtually anyone can take a stab and, ultimately, solve this problem with a high degree of precision and confidence. And this case competition aimed at proving this point.

‍

‍

The Setup  

Students received uncleaned training data set and a validation dataset of factors like location, area type, size, society, number of rooms, balcony, prices. For the submission, students had to create a data app to predict the prices of houses provided in the prediction dataset and explain the working of the data app in a video.  

‍

The second step in the case competition setup was a brief 1.5-hour training workshop to kick off the competition so that students can get familiarised with the Ikigai platform. As a follow-up, students received access to the training resources, product documentation and joined Ikigai’s Discord community. To help the participants navigate the challenge, the Ikigai support team organized weekly office hours. “We were deeply impressed by the enthusiastic participation in the competition. The students became proficient users in just a few weeks,” says Vinay Gahlot, the case competition lead.

The Winners and Their Solutions  

By using Machine Learning techniques, students identified trends and patterns in the housing market, and developed solutions to predict the prices of houses in less than four weeks. Some of the teams removed the outliers from the data to gain better prediction accuracy and developed their ML models apart from the inbuilt models provided by Ikigai. A few key solutions developed prediction models for both buyers and sellers, presented solutions on geolocation maps and did in-depth exploratory data analysis.

The overall best solution was provided by team Paper Canary (Khushi Solanki, Shubh Gajjar, Pragnesh Barik) which used the XGboost model to provide the most accurate price prediction and developed a data app for both buyers and sellers. These two factors made them stand out and got team Paper Canary first place in this case competition. Link to the submission of team Paper Canary: https://youtu.be/K5UlwgAf_zs  

Team KUINE (Atharv Agarwal, Nayan Lakade, Shine Priyan) emerged as the runner-up in this competition as they put great efforts into cleaning the data type using phonetic clustering, changing data types and removing outliers before the prediction of house prices using regression and clustering. Moreover, team KUINE prepared a beautiful dashboard consisting of visualisations of price vs availability and a heat map of price/area to help users in choosing which areas to invest in. Solution link of team KUINE: https://youtu.be/wXjaFzN5qz8

‍

Team Qwerty (Nishant Kumar Singh, Punit Kumar Mishra, Harshit Bhardwaj) secured the third position in this case competition. They performed good exploratory analysis and did outlier removal based on Gaussian distribution. Also, they used liner regression, random forest and ridge regression ML models to train the dataset for predicting house prices. Link to the solution developed by team Qwerty: https://youtu.be/OMPYzZAPDZc

‍

More creativity, less coding

The participants had a smooth experience using the platform. They find it easy to understand for new users. A few quotes about the competition and platform: "It was a very well-organized and engaging competition that allowed me and my team to explore new methods to tackle problems. The platform provided us with a great interface to use tools and reduce the amount of coding effort”- Nishant Kumar Singh, “Got to explore different data processing techniques in this competition and the platform had a very straightforward and intuitive user interface”- Shine Priayn.

No-code AI is a fast way to build solutions without needing additional training. Participants appreciated how the no-code/low-code solutions significantly reduce software development time. Using tools like Ikigai provides more time and opportunity to be creative and to think about the underlying business logic rather than tackling ML, coding, and implementing, which increases the speed of problem solving. It allows users to rapidly develop new solutions to transform business processes and meet ever-changing customer needs.  

‍

Closing Thoughts

Case competitions like this provide a valuable learning opportunity for students. By working on real-world problems and using data analytics to develop solutions, students are able to gain valuable experience and skills that will be beneficial in their future careers.  

This competition helped students in developing skills like outlier removal, data visualisation, machine learning, time management, project management, teamwork, and computational thinking. Our participant Sine Priyan said that “I used the geographical representation of data and I feel that it gives a very insightful visualization and I will be using it more to represent data”. Aditya Yadav mentioned that “I am going to make an ML project in near future this competition surely makes it a lot easier for me and give me confidence for it.”

Overall, the use of data analytics in case competitions to tackle the Bangalore housing problem is a great example of how college students can use their skills to make a positive impact in their community.

‍