Data Mining @UPM2021

There are three datasets involved in the project, namely LRGS TUA, KWAP and IEEE CASSMiLe. All dataset are based on real problem, and challenge based learning approach is used to complete the learning outcomes.

PROJECT DATASET 1: LRGS TUA

One of the datasets being used is called LRGS Tua: Neuroprotective Model for Longevity Among The Malaysian Elderly, a national project sponsored by the Ministry of Higher Education, Malaysia. The dataset used for this project is provided by the Malaysian Research Institute On Ageing (MyAgeing™), Universiti Putra Malaysia. The dataset contains indicators to mental state and geriatric depression of elderly from four states. The attributes are:

Demographic - age, gender, state, main job, job sector, marital status and live with status

Health - smoking, alcohol intake, MMSE level, Mini-Mental State Examination (MMSE), activities on daily living (ADL) and WHODAS baseline (disability)

Social - Lubben social network, neighborhood, total social cohesion scale, medical outcome study social support (informational), medical outcome study social support (tangible support), medical outcome study social support (affective support), medical outcome study social support (positive social interaction)

Psychology - quality of life, total SWLS, EPQ average, total loneliness, flourishing scale score and Geriatric Depression Scale (GDS-15)

Group: LRGS1

This all girls group produced a prediction tool that can detect dementia on elderly based on an indicator; MMSE (Mini–Mental State Examination) that is able to measure cognitive impairment. They conducted the following experiments:

Experiment 1 : To predict MMSE based on original dataset from domain of psychology and health

Experiment 2 : To predict MMSE based on removed MMSE-Normal dataset from domain of psychology and health

Experiment 3 : To predict MMSE based on removed MMSE-Normal dataset from domain of psychology and health that is resampled

https://sites.google.com/student.upm.edu.my/e-portfolio200891/home

https://sites.google.com/view/lrgsproject-200471/home?authuser=0

https://sites.google.com/student.upm.edu.my/lrgs-tua-200456/introduction

https://sites.google.com/student.upm.edu.my/my-e-portfolio/home

DATA MINING ONLINE PORTFOLIO – Presentation (16:9) by Najma Hassan

The objectives of this group are:

To predict how all the factors from Demographic, Social, Social-economic and Health affect the Mini Mental State Examination result
To predict how all the factors from Demographic, Social, Social-economic and Health affect the Geriatric Depression Scale result

Group: LRGS2

A group consists of Najma, Abida, Aiman, Wahaj and Pujeeta developed two predictive models, one using MMSE (mini mental state examination result: severe/normal/mild) and another for GDS (geriatric depression scale: depress/normal). RapidMiner and Python are used to develop Decision Tree and Naive Bayes models. They performed resampling and PCA for pre-processing, and hyper parameter tuning.

https://sites.google.com/view/pujeetaasri202331/home

https://sites.google.com/view/najma-195773/home?authuser=0

https://sites.google.com/view/abida-199714/home?authuser=0

https://sites.google.com/view/ssk4604-data-mining-aiman/home?authuser=0

https://sites.google.com/view/wahajsiddiqui-201439

Group: LRGS3

Ths group focused on clustering based analysis and develop models for prediction on MMSE Result of LRGS Tua Dataset by 3 experiments:

Exp 1: LRGS (Cleaning) dataset

Exp 2: LRGS Resample dataset

Exp 3: LRGS Resample + Hyperparameter Tuning dataset

to develop a greater understanding of how the elderly living their life in the age of 60s and above and to analyze the reason behind of it as well as to observe how different categories (Health, Psychology, Social, Economy and Demographic) can affect the MMSE result of the respondents.

https://sites.google.com/view/e-portfolio-afiqah202872

https://sites.google.com/view/200714e-portfolio/home?authuser=0

https://sites.google.com/view/lrgs-tua/home

https://sites.google.com/student.upm.edu.my/ssk4604-datamining/home

Data Mining Presentation – Presentation (16:9) by nunu

PROJECT DATASET 2: KWAP

This dataset is provided by the provided by the Malaysian Research Institute On Ageing (MyAgeing™), Universiti Putra Malaysia based on one of their past projects for Kumpulan Wang Amanah Persaraan (KWAP), the Malaysia’s largest public services pension fund. The dataset consists of several attributes such as age, marital status, level of education, socio economic, salary, attitude towards retirement, self-rate health, financial well being, self care, life satisfaction, f2healthstat, and Level distress. The dataset can be used to predict the risk of health and finance for retirees-to-be.

Group: KWAP1

This group develop two predictive models and two Streamlit application; one for financial distress prediction and another is for health risks. They conducted 3 experiments to compare the accuracies. The first was using dataset that has been resampled. The second was using dataset that has been resampled and standardized. The last experiment was using dataset that has been resampled, standardized and discretized. They also used 2 different training and testing ratio which are 80:20 and 60:40.

https://sites.google.com/student.upm.edu.my/kwap-project-portfolio/home

https://sites.google.com/student.upm.edu.my/kwap-project-data-mining/home

https://sites.google.com/student.upm.edu.my/datamining-kwapproject-adlina/intro

https://sites.google.com/student.upm.edu.my/dataminingproject/home

Group: KWAP2

This group focused on the optimization of the Random Forest and Decision Tree through experiment that investigates the maximum depth by altering the number to 5, 10, and 25.

https://colab.research.google.com/drive/1z3JWPt6lW52-Q2mQyHJBqW1iStYQHXdb?usp=sharing#scrollTo=VM2Xn3Fvk5M6

https://sites.google.com/student.upm.edu.my/eportfolio-202876/home?authuser=0

https://sites.google.com/view/kwapnorhidayati/homepage?authuser=0

PROJECT DATASET 3: CASSMiLe Logs

Only 1 group has focused on the microcredential logs provided by CASSMiLe, an online learning initiative by IEEE Circuit and Signal System society. The dataset contains access logs by CASSMiLe's global learners to each of the offered lesson. The analytics focus on the lesson's achievement, learners participation, and badges completion.

This duo has came put excellent effort to build the dashboard for lesson and learners analytics using PowerBI. The predictive model focused on who will complete the lesson. The performed resampling and hyperparameter tuning using GridSearch and tree depth to obtain the best model.

https://sites.google.com/student.upm.edu.my/syafi-eportfolio/home?authuser=0

https://sites.google.com/view/hezzarry-eportfolio/home?authuser=0

For more information, contact nurfadhlina@upm.edu.my

Google Sites

Report abuse