There are 0 repository under wids-datathon topic.
WiDS Datathon 2020 on patient health through data from MIT’s GOSSIS (Global Open Source Severity of Illness Score) initiative.
Predicting the presence of an oil palm plantation in an image
This repository encompasses my submission for the WIDS (Women In Data Science) Challenge for 2018, held in Jan-Mar'18. It was an invite only competition based on InterMedia Survey Institute, a grant recipient of the Bill & Melinda Gates foundation in their Financial Services for the Poor program. Participants were required to predict gender based on demographic and behavioral information of survey respondents from India and their usage of traditional financial and mobile financial services.
The dataset was created in collaboration with Climate Change AI (CCAI). Participants will submit forecasts of temperature and precipitation for one year, competing against the other teams as well as official forecasts from NOAA.
WiDS Datathon 2023 Contest 13th place WiDS Global and 1st place WiDS Hanoi solution by UET 杨蓉 Team (Binh Nguyen Thai, Duc Tran Van, Duong Tran Thuy, Dung Tran Thuy)
combination of EvalML with Rapids for the WiDS 2021 competition
Using data within first 24 hours of intensive care to develop a machine learning model that could improve the current patient survival probability prediction system (apache_4a) and is more generalized to patients outside of the US
Predicting Diabetes Mellitus in ICU patients with data from the patient's first 24 hours.
Data Science practice for the WiDS 2022 Datathon
Diabetes melitus adalah penyakit kronis yang terjadi ketika pankreas tidak dapat menghasilkan hormon yang mengatur gula darah (insulin) dengan cukup, atau ketika tubuh tidak dapat menggunakan insulin yang dihasilkannya. Prevalensi diabetes meningkat lebih cepat di negara berpenghasilan menengah dan rendah. Diperkirakan 1,6 juta kematian akibat diabetes dan darah tinggi sebelum usia 70 tahun. Diabetes dapat diketahui dengan cara tes kadar gula dalam darah dan check-up. Namun, karena keterbatasan waktu dan tenaga, tidak dapat dilakukan check-up kepada seluruh pasien. Penelitian sebelumnya telah banyak yang menggunakan machine learning untuk mendiagnosa penyakit diabetes, namun belum ada penelitian yang menggunakan data 24 jam perawatan pertama pasien ICU. Pada penelitian ini penulis menggunakkan data 24 jam pertama perawatan pasien ICU. Penulis menggunakan teknik data cleaning, fill missing value dan One Hot Encoding pada preproccesing. Penulis membandingkan algoritma LightGBM, XGBoost dan CatBoost dengan hasil akurasi, AUC, Precision, Recal, F1 Score sebesar 83.73%, 86.45%, 67.35%, 48.11% dan 56.13%, kemudian melakukan hyperparameter tuning kepada algoritma CatBoost dan mendapatkan hasil akurasi, AUC, Precision, Recal, F1 Score sebesar 83.72%, 86.60%, 67.26%, 48.24%, 56.18%.