Novel Approach to Salmon Stock Prediction: Machine Learning and Deep Learning algorithms
Thomas Yuan
St George's School
Floor Location : S 041 V
Right before summer, our school went to a remote town in BC called Cheakamus, where we studied the environment there. Cheakamus is dealing with some data problems that can be solved with computer engineering. In Cheakamus, students get to record data such as the river temperature, precipitation, and water. However, every year hundreds and thousands of students go to Cheakamus to learn and study, yet there is not a convenient way to learn the environmental data collected. More specifically, students collect lots of data but are not able to understand the relationship between the different collected data. However, studying the salmon population is a highlight in Cheakamus. Many students enjoy this process, and genuinely care about the salmon population. Thus, it is crucial to know how to preserve the population or why it was not preserved. Through experiments, I found that the data that students collect is highly correlated with the salmon population. Therefore, a salmon forecasting software was created.
The software that is created to solve this problem is a statistical analysis software that uses machine learning to analyze the relationship between variables and predict salmon spawning. In the current market, there are many softwares that are designed to be utilized by companies with huge data sets. Due to the complexity of the statistical analysis softwares, these softwares are not useful to students who are trying to understand the environment. In addition, the current statistical analysis apps are not specific to analyzing and predicting the salmon population. Therefore, even if students could use the current statistical analysis software, they would not get an accurate prediction of salmon spawning.