“Artificial Intelligence, deep learning, machine learning — whatever you’re doing if you don’t understand it — learn it. Because otherwise, you’re going to be a dinosaur within 3 years,” said Mark Cuban, billionaire technology entrepreneur, investor and TV personality. How, you ask? Through practice, of course. Learn the basics and then spend a lot of time practising your skills on machine learning projects. Allow us to give you some inspiration and get you started.

Machine learning Projects to be Inspired by

Machine learning applications are everywhere. From banks to bars, ML is changing the way people understand the past, live the present and plan the future. Here are some of the best machine learning projects we’ve seen.

Machine Learning Projects: Classification and Recommendation Projects

#1 Netflix recommendation

No list about machine learning projects can exist without talking about Netflix’s world-renowned work on their content recommendation engine. In addition to your own viewing history and search queries, Netflix correlates data with users with similar interests. Read about how Netflix works. Go ahead and build your own movie recommendation engine with these machine learning datasets.

#2 Uber Eats food discovery

Uber Eats’ food recommendation engine uses Graph Neural Network (GNN) to recommend food items. This happens in two phases: 

  1. It generates dishes and restaurants based on pre-filtering such as distance from the user etc.
  2. It then adds user personalization based on past queries, orders etc. and contextual information like weather, time of day, current location etc.

Uber engineers write about this project in detail here.

#3 Gmail spam filter

The spam filter in Gmail is one of the most advanced classification mechanisms in the world. In fact, an email marketer’s biggest challenge is circumventing the spam filter, who works overtime to find sneaky ways. Last year, Google made their protections even stronger, powered by the machine learning tools of Tensor Flow, which identifies and eliminates sneaky spam. Read about the project here.

Machine Learning Projects: Prediction and Forecasting Projects

#4 Walmart sales forecasting competition for hiring

Walmart announced a machine learning model-building competition on Kaggle for ML engineers aspiring to join their teams. They opened up sales data for 45 stores over a two-year period and invited job-seekers to forecast future sales. Here are the machine learning datasets for you to try your skills on. In case you need help, here is a step-by-step guide for forecasting using various algorithms like linear regression, neural network and Facebook’s Prophet etc. 

#5 StatWorx stock prices forecasting hackathon

A European AI/ML consulting company called StatWorx conducted a hackathon where they scraped data from S&P 500, a US stock market index. They then used this data to predict market trends in the future using deep learning algorithms. Here’s their story and step-by-step guide. Here’s a similar data set for India’s National Stock Exchange. Run your own machine learning case studies on this data set.

#6 House price prediction

Real estate prices are often viewed as an indicator of the state of the economy. But a lot of factors go into gauging the price of a house — size and carpet area, the crime rate in the neighborhood, school quality etc. Here’s a look at how you can predict house prices by building a correlation model, simple and multivariate regression.

Machine Learning Projects: NLP, Image Processing Projects

#7 Handwriting recognition in Malayalam

Researchers from Kochi, Kerala recently presented a complex and challenging project on handwriting recognition in Malayalam manuscripts. They used contour detection for segmentation and convolutional neural network (CNN) for classification. You can build your own project using this ML data set. If Malayalam is not your language, here are some data sets in other Indian languages as well such as Tamil, Kannada, Hindi (Devanagari) and Bengali.

#8 Stanford’s movie review sentiment analysis project

A typical sentiment analysis algorithm analyses word by word or at best phrase by phrase. This produces unsatisfactory results. So, Stanford researchers used recursive neural networks and sentiment treebank to improve the model. You can read about the project here. You can try your skills on the dataset here.

#9 Speech emotion recognition

Knowing how someone feels is tough even when you’re in a face-to-face conversation. Imagine how difficult it will be to identify human emotions only through their audio recording! This cool project uses deep learning techniques to find out the emotion of the person from their speech recording. This will come to be highly useful in training contact center/help desk representatives very soon. 

#10 License plate recognition

Detecting and identifying information from moving vehicles are interesting machine learning case studies for image recognition professionals. One, because it’s challenging. More so, because it has a wide-ranging impact across use cases. Here’s a cool implementation of this machine learning project using YOLO (You Only Look Once), an up and coming object detection library. First, the object (i.e. the license plate) was detected from the image. Then, the characters were extracted and segmented. Finally, it compared the record against existing systems and recognize the vehicle. 

#11 Machine Learning Projects: Make your own

You can read about a million projects from around the world. But, if you want to become a professional, you need to do your own work. So, take inspiration from the above, pick a data set and start your own machine learning project. If you’re not ready for practical projects yet, consider Springboard’s 1:1 mentor-led project-based machine learning program to build your foundation.