NUST Institutional Repository

Smart pollen monitoring and prediction service

Show simple item record

dc.contributor.author Samad, Abdus
dc.contributor.author Farooq, Saqib
dc.contributor.author Supervised by Asst. Prof. Dr. Saddaf Rubab
dc.date.accessioned 2020-11-13T05:53:48Z
dc.date.available 2020-11-13T05:53:48Z
dc.date.issued 2020-07
dc.identifier.other PCS-390
dc.identifier.other BESE-22
dc.identifier.uri http://10.250.8.41:8080/xmlui/handle/123456789/11630
dc.description.abstract Technological developments have revolutionized how we deal with many problems in our lives. We can find out about a lot of things before they happen thanks to technologies like Artificial Intelligence and Machine Learning. Many types of allergies can be avoided by taking some precautionary measures like avoiding a specific food or drug. Pollen allergy is hard to deal with because its windborne as well. This is what we are trying to solve. We want to make pollen patients aware of any outbreak or spike of pollen density before it even happens so they can take precautionary measures. But is it possible? What if we can look into the future and tell exactly how severe is the pollen going to be tomorrow? or the day after that? Or next week? However impossible it may sound, Thanks to Machine Learning we did just that. Using a huge dataset that contains data for different attributes like the count of pollen, humidity and temperature of every single day for the past 10 years, we built multiple ML models on which we trained this dataset. The models when trained were able to detect patterns in the data. These patterns could now be used to predict the density of pollen in future based on the training data set. We used three Machine Learning models of which Logistic Regression gave us the best accuracy of 99%. Being this accurate isn’t very likely when dealing with other real world problems but because our dataset was very linear and the patterns were fairly visible, i.e the spike in pollen count in spring season specifically, we could get the accuracy as high as 99%. The statistics we get from the Machine Learning algorithm will be displayed on an android application that we built from scratch. The user will have to install the android application and select the type of pollen allergy and his area. The application will show him if the pollen density is low, high, severe or critical for the next few days. The user can also set the notifications for when the density of pollen is alarming. This will help him take precautionary measures and stay inside in case there’s an outbr. en_US
dc.language.iso en en_US
dc.publisher MCS en_US
dc.title Smart pollen monitoring and prediction service en_US
dc.type Technical Report en_US


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account