Thursday 06 March 2025
The quest for accurate house price predictions has long been a challenge in Ghana, where the rental market is notoriously opaque and complex. But now, thanks to the power of machine learning, researchers have developed a model that can accurately forecast prices based on a range of factors.
The study, which drew on data from over 17,000 property listings, used CatBoost, an ensemble-based algorithm, to identify the key drivers of rental prices in Ghana. The results were striking: location, number of bedrooms and bathrooms, and amenities such as air conditioning and refrigerators emerged as the most significant factors influencing price.
But what’s perhaps most impressive is the accuracy of the model itself. By training CatBoost on a dataset of property listings, researchers were able to achieve an R2 score of 0.877 – a measure of how well the model fits the data. This means that for every dollar in actual rental prices, the model was able to predict with remarkable precision.
The implications are significant. For renters and landlords alike, having access to accurate and reliable house price predictions could be a game-changer. No longer would they have to rely on guesswork or anecdotal evidence when making decisions about renting or selling property.
But this study is more than just a useful tool for individual stakeholders – it also has broader implications for the Ghanaian economy as a whole. Accurate house price predictions could help policymakers make informed decisions about issues like housing affordability and supply, ultimately shaping the country’s economic landscape.
So how does CatBoost work its magic? Essentially, it uses a combination of decision trees to identify patterns in the data that are not immediately apparent. By combining multiple trees, the algorithm is able to capture complex relationships between variables and make more accurate predictions.
The model was trained on a dataset that included factors such as property type, location, size, and amenities. Researchers also included demographic information about the properties’ owners and tenants, as well as data on local economic conditions.
One of the most interesting findings was the importance of location in determining rental prices. Properties located in areas with high demand – think desirable neighborhoods or proximity to public transportation hubs – tended to command higher prices. Conversely, those in less desirable areas saw their values drop.
Amenities also played a significant role, with features like air conditioning and refrigerators pushing up prices. This makes sense, given the premium that many renters place on comfort and convenience.
Cite this article: “Accurate House Price Predictions in Ghana Using Machine Learning”, The Science Archive, 2025.
Machine Learning, Ghana, House Price Predictions, Rental Market, Property Listings, Catboost Algorithm, R2 Score, Accurate Predictions, Housing Affordability, Economic Landscape, Decision Trees, Demographic Information, Local Economic Conditions.
Reference: Philip Adzanoukpe, “Predicting House Rental Prices in Ghana Using Machine Learning” (2025).
