EazyML Blog

Easy To Use, Easy To Learn

EazyHacks: EazyML Hackathon

November 19, 2020

Sai Vedagiri, Samuel Hampel, Xavier McLeod, Raymond Shen

High Technology High School

EazyML Team Thanks everyone who participated in the Hackathon. A special Thanks to the Judges: Dr. Courtney Paulson, Professor in Business Analytics at Southern Utah University and Mr. Promod Radhakrishnan, Global Vice President, Oracle (Financial Services Global Business Unit).

EazyHacks Hackathon took place Friday, November 13 to Sunday, November 15. This hackathon was hosted by EazyML, where participants had 40 hours to create an application from scratch, incorporating the use of machine learning by utilizing the platform or API provided by EazyML.

Participants were allowed to find their own datasets to run the machine learning algorithm, but data on the spread of California wildfires was also provided. California has always had wildfires which have some positive effects such as increasing nutrient density in the ground and promoting new environmental growth. In recent years however, the wildfires in California have gotten worse at an extremely high rate. According to the California Department of Forestry and Fire Protection, as of November 1, 20201, almost 10,000 wildfires have caused over 4 million acres of forest to be burned, which is almost 16 times the acreage burned from the same time period last year. Many factors go into figuring out when and where these fires will form and how deadly they could be. Important predictors that point to an influx of wildfires include a lack of precipitation, high wind speeds, high temperatures, and a lack of forest management per location. By analyzing these factors along with other pertinent information, machine learning can be used to make predictions so that the Californian government can allocate the correct amount of resources to the correct locations to deal with wildfires in a timely manner with minimal waste.

During the hackathon, some participants used provided data of the California wildfires to create predictions on the fires. One such project included analyzing the fires and applying it to people who are considering moving to California. If the fires are projected to increase significantly in a certain county or area, then the person using the application would be advised to reconsider going to that area. Another project used the same data in a similar way, but applied it to people who were already living in areas of California.

While many participants tackled the severe problem of the California wildfires, other participants opted to do other exciting topics that can make an impact. One such group created a bot that analyzes tweets about a certain movie. By analyzing them by topic and sentiment of the movie, the tweets can be categorized and displayed by category, allowing someone to find out what others generally thought about each movie. Another group tried to tackle the issue of global warming, by tracking glaciers and predicting where they will be in the future.

In all of the projects described above, EazyML played a crucial role in the back-end processes. The machine learning API provided by EazyML was integrated into python programs written with all of the projects submitted. Most involved plugging in a simple csv file to train the data, then introducing another spreadsheet that will be used for finding the final data. EazyML’s simple process for generating predictions and results, along with the many different options for building models meant that even with the same dataset, a variety of different outcomes were produced from each project submitted and were displayed in different ways. One project even showed a comparison between EazyML and their own generated neural network, in which EazyML’s superior performance with things like bias and outlier detection led to the EazyML generated results being used over the other results.

Although all of the projects submitted were flushed out ideas which skillfully integrated EazyML’s machine learning with a real world application, some of the projects stood out much more than the others, based on creativity, complexity of design, and the use of EazyML within the application.

The project that took first place, winning a variety of prizes, was the Glacier Time-Machine. The group who made it wanted to raise awareness to the rapidly changing environment due to global warming, and they were able to use EazyML to predict the sizes and locations of glaciers in the future. By arranging the past, present, and future places of glaciers in the world, people are able to see how severe the effect of global warming has been on the oceans on Earth. The website created also provides a lot of information and many useful tips on global warming and how anyone can help mitigate climate change.

Here is a link to the 1st Place winner EazyML project:

The project that placed second was California Wildfire Intensity Explanation. By tweaking with EazyML’s API and the many models, the group behind the project was able to figure out the best model that would most accurately predict the intensity of California wildfires in different areas. They then provided an accurate and concise explanation of how the machine learning algorithm came to its conclusion, which not only helps in predicting any place that may suffer from California wildfires, but will also empower other people who come across this project to understand how the fires may be spreading and what they can do to stop this.

Here is a link to the 2nd place winner EazyML project:

The project which earned third place, FireSense, also focused on California wildfires, but took a different approach than the second place project. Instead of taking in data from past wildfires and where they were, this group took a more proactive stance towards immediate detection through the use of images. By setting up a camera attached to a rotating servo motor, pictures are able to be intaken that scans the surrounding area. In addition to having cameras out in nature that can detect any fire that is actively raging near it, alerting any firefighters earlier than other methods might be able to, a system attached to the camera stores the image data for future use. Pixel color values from these images are sent into EazyML, and the machine learning model provides a prediction of which areas are likely to catch fire.

Here is a link to the 3rd place winner EazyML project:

While many ideas and projects were made into reality during the length of EazyHacks, there are still many applications of EazyML’s machine learning platform that have not yet been realized. Data is found everywhere around the world and on the internet, and once that data is collected, there is an infinite number of uses for it. If one would like to make predictions with large and complex data, EazyML can help . The platform has an easy to understand API which can be integrated smoothly into any python program that needs machine learning.EazyML is able to provide students, professionals, or even hobbyists the models necessary for analyzing and portraying many aspects of our natural society. With just some data and EazyML, you can predict your world.

You can get started with EazyML on https://eazyml.com or https://eazyml.com/docs as a developer for building on the API. Our next Hackathon is scheduled for November 29th. For more information or to register, please visit: https://hacktable2020.devpost.com/ If you have more ideas or want to reach us for Hackathon sponsorships, please reach us at info@eazyml.com

Share This Post