The package implements two iterative techniques called
T3Clus and 3Fkmeans, aimed at simultaneously clustering objects and a
factorial dimensionality reduction of variables and occasions on three-mode
datasets developed by
\(\href{https://doi.org/10.1007/s00357-007-0006-x}{\text{Vichi et. al}}\) in
2007. Also, we provide a convex combination of these two simultaneous
procedures called CT3Clus and based on a hyperparameter \(\alpha\) (\(\alpha
\in [0,1]\), with 3FKMeans for \(\alpha=0\) and T3Clus for \(\alpha=1\)) was
also developed in
\(\href{https://doi.org/10.1007/s00357-007-0006-x}{\text{Vichi et. al}}\).
Furthermore, we implemented the traditional tandem procedures of T3Clus
(TWCFTA) and 3FKMeans (TWFCTA) for sequential clustering-factorial
decomposition (TWCFTA), and vice-versa (TWFCTA) proposed by
\(\href{https://doi.org/10.1007/978-3-642-79999-0_1}{\text{P. Arabie and L.
Hubert}}\) in 1996.
| source codes | publication |
|---|---|
| Python (Github) | PyPI |
| R (Github) | R-CRAN |
As the creator of The Addon Forge, I design and develop high‑performance browser extensions focused on web automation, productivity enhancement, and media extraction. The brand emphasizes clean engineering, user‑centered design, and seamless cross‑browser compatibility.
My published extensions enable efficient downloading of product videos and media assets from major e‑commerce platforms such as Alibaba and AliExpress, with upcoming releases expanding into YouTube media extraction and additional workflow optimization tools. All extensions are actively maintained, performance‑optimized, and built using modern JavaScript and WebExtension APIs.
| Source Codes | Publication |
|---|---|
| Alibaba Media Downloader |
Chrome Web Store Firefox Add‑ons |
| AliExpress Media Downloader |
Chrome Web Store Firefox Add‑ons |
| The Addon Forge YouTube Channel | Tutorials, Demonstrations, Release Notes |
| YouTube Music Video Downloader | Upcoming Release |
In this study, we conducted a case
study on improving the National Health Insurance Scheme (NHIS) coverage in the
Sunyani municipality. We used the Contingent Valuation Method (CVM), a technique for
estimating the Willingness To Pay (WTP) for non-market goods, to determine
individuals' willingness to pay for an extension of NHIS coverage. We also employed
Binary Logit Regression analysis to identify factors that influence individuals'
decisions to pay for an improved NHIS, and included a social interaction term in the
model to measure the impact of other individuals' views on an individual's decision.
Our research results showed that a majority of NHIS holders who purchase additional
drugs when visiting a health facility were willing to pay for an extension of NHIS
coverage. The most costly medical conditions for which respondents usually purchased
additional drugs and were willing to pay an additional premium for inclusion in NHIS
coverage were rehabilitation conditions (such as vision, hearing, and dental
issues), cancer, and heart-related surgeries. On average, respondents were willing
to pay no more than GH₵30.00 per year to cover these conditions. We recommended that
policymakers or planners consider income, gender, and level of education when making
NHIS coverage improvement decisions, as these variables were found to be the most
significant in our model.
In this work, we developed and
implemented five models, TWCFTA, TWFCTA, T3Clus, 3FKMeans, and CT3Clus, in the
simuclustfactor packages in Python and R for the analysis of three-mode datasets.
These models were proposed by
\(\href{https://doi.org/10.1007/s00357-007-0006-x}{\text{Vichi et al.}}\) in 2007 as
an alternative to traditional tandem models for simultaneously clustering objects
and reducing the dimensionality of variables and occasions in three-mode datasets.
T3Clus and 3Fkmeans are iterative techniques that apply the Tucker2 algorithm and
K-means algorithm sequentially, while CT3Clus is a convex combination of these two
methods with a hyperparameter alpha \((0 \leq \alpha \leq 1)\) that allows for the
interpolation between the two. In contrast, tandem analysis only involves the
sequential application of clustering and factorial methodologies. We tested these
simultaneous models on synthetic and real datasets and found that they produced more
well-separated clusters with higher cohesion within clusters, and were better at
recovering object-cluster, variable/occasion-factor associations compared to the
tandem models. The best results in terms of well-separated clusters with high
cohesion within the clusters were obtained when the factorial analysis was
prioritized, minimizing the within-cluster deviation of the component scores (i.e.
WSS in the reduced space).
University of Aveiro
(Aveiro, Portugal)
MSc. Mathematics & Applications
August 2021 - Sept. 2022
core courses :
supervisor:
University of L'Aquila
(L'Aquila, Italy)
MSc. Mathematical Engineering
July 2020 - July 2021
core courses :
supervisor:
University of
Energy and Natural Resources
(Sunyani, Ghana)
BSc. Actuarial Science
August 2019 - September 2020
core courses:
supervisor:
Achimota Senior High
(Accra, Ghana)
General Science
August 2011 - September 2014
core courses:
BNP Paribas – Marketing Data Analytics Hub
(Porto, Portugal)
Python Developer / Internal Tools Engineer
March 2023 – March 2025
core activities:
Xtratek Automation
(Douala, Cameroon)
Data Scientist
2020 – 2022
core activities:
Python Software Foundation (PSF)
(Sunyani, Ghana)
Team Lead
June 2019 - 2021
core activities:
University of Energy & Natural Resources
(Sunyani, Ghana)
Teaching Assistant
September 2019 - September 2020
core activities:
New Fountain International School
(Tema, Ghana)
Pupils Tutor
2014 - 2015
core activities:
IFCS 2022
(University of Porto, Portugal)
17th Conference of the International Federation of Classification Societies
July 2022
Contributed session
Machine Learning in Science
(University of
Lisbon, Portugal)
Workshop of Machine Learning in Science
June 2022
participant
core topics:
core topics:
core topics:
core topics:
core topics:
case study of AI in:
Outline:
Outline:
