Ishango
Africa's Only Fully Funded Data Science Fellowship Program
Case Studies
Ishango.ai data scientists work remotely on value-adding projects for international companies. Our data scientists have delivered projects across sectors such as finance, e-commerce, healthcare, agriculture, and industrial engineering, for companies from around the world including the US, Australia, Switzerland, and the UK.
Case Studies
As part of our two month data science programme, Ishango.ai data scientists work remotely on value-adding projects for international companies. Our data scientists have delivered projects across sectors such as finance, e-commerce, healthcare, agriculture and industrial engineering, for companies from around the world including the US, Australia, Switzerland and the UK.
Our Projects
From remote sensing and building recommender engines to using machine learning to predict disease patterns, Ishango.ai data scientists have worked on real-world, high-impact projects that created value for their host organisations. Read our case studies below to learn more about our work.
Our Case Studies
Project: To better understand the impact of demographics on mental illness by understanding the relationship between the demographic variables, the disease, and the connectomic data.
Outcome: Joseph Domguia and Anisie Uwimana were able to build a model which successfully identified key features, such as age and sex which contribute to the predictions of their model.
Project: To analyse how Simprint’s fingerprint technology was being used and its effectiveness as a biometric tool for identification.
Outcome: The analysis that Victoire Djimna delivered enabled Simprints to evaluate how their technology was performing, to make improvements to project operations on the ground, and also to influence the design of future projects.
Project: To develop an unsupervised machine learning model to predict when parts are likely to fail within a hydroelectric power plant.
Outcome: Faith Benson and Cyrille Feudjio developed a model that could identify anomalies at certain times and also identify abnormal patterns which are likely to be the cause of the anomalies detected.
Project: To develop a system to remotely identify tobacco fields using satellite imagery data aiming to improve the socio-economic well-being of the farmers by suggesting best practices.
Outcome: Bright Aboh and Aimable Ishimwe Manzi developed a model that could identify tobacco leaves and therefore tobacco farms and also measure the land surface on which the crop was being grown on.
Project: To design a data-driven system that could create targeted recommendations of suitable products to tails.com customers, aiming to introduce new products to customers.
Outcome: Sylvera Massawe and Aurelie Jodelle Kemme developed a recommender system targeting products to pre-existing tails.com customers.
Project: To develop a recommendation engine to support Analysts by suggesting high-relevance articles as a source of insights for their clients.
Outcome: Stephen Adjignon and Winifred Harriet Asante developed an NLP model to extract and group topics using Latent Dirichlet Allocation (LDA) topic modelling
Project: To reduce the time Phastar analysts spend manually coding verbatim medical terms by developing an automated Natural Language Processing (NLP) solution to identify and recommend the most similar Low-Level Terms (LLTs) for every verbatim term.
Outcome: Samuel Nignamon Gyimah and Opanin Agyei Adu developed an NLP model that could identify and recommend the most similar LLTs with up to 90% accuracy
Project: To identify inefficiencies in production lines and extract actionable insights from the data provided.
Outcome: Development of a non-linear supervised machine learning approach (random forest) to better explain variables affecting productions and detect abnormalities at large scale
Project: To develop a python-based model to estimate the size of pineapples from phone-generated imaging directly taken from plants in the field and to apply this model to internally developed correlations to predict pictures from pineapples into sizes.
Outcome: Hundreds of pineapple images were annotated to train the deep-learning model, which achieved up to 90% accuracy. Reasons for the remaining errors as well as potential solutions were identified.
Get the Case Study
Privacy Policy | Copyright © Ishango 2023