The numerical value of k in a k-fold cross-validation training technique of machine learning predictive models is an essential element that impacts the model’s performance. A right choice of k results in better accuracy, while a poorly chosen value for k might affect the model’s performance. In literature, the most commonly used values of k are five (5) or ten (10), as these two values are believed to give test error rate estimates that suffer neither from extremely high bias nor very high variance. However, there is no formal rule. To the best of our knowledge, few experimental studies attempted to investigate the effect of diverse k values in training different machine learning models. This paper empirically analyses the prevalence and effect of distinct k values (3, 5, 7, 10, 15 and 20) on the validation performance of four well-known machine learning algorithms (Gradient Boosting Machine (GBM), Logistic Regression (LR), Decision Tree (DT) and K-Nearest Neighbours (KNN)). It was observed that the value of k and model validation performance differ from one machine-learning algorithm to another for the same classification task. However, our empirical suggest that k = 7 offers a slight increase in validations accuracy and area under the curve measure with lesser computational complexity than k = 10 across most MLA. We discuss in detail the study outcomes and outline some guidelines for beginners in the machine learning field in selecting the best k value and machine learning algorithm for a given task.

Skin diseases are reported to be the most common disease in humans among all age groups and a significant root of infection in sub-Saharan Africa. The diagnosis of skin diseases using conventional approaches involves several tests. Due to this, the diagnosis process is seen to be intensely laborious, time-consuming and requires an extensive understanding of the domain. The enhancement of computer vision through artificial intelligence has led to a more straightforward and quicker way of detecting patterns in images, which can be harnessed to equip diagnosis process. Despite the breakthrough in technology, the dermatological process in Ghana is yet to be automated, making the diagnosis process complicated and time-consuming. Hence, this study sought to propose a web-based skin disease detection system named medilab-plus using a convolutional neural network classifier built upon the Tensorflow framework for detecting (atopic dermatitis, acne vulgaris, and scabies) skin diseases. Experimental results of the proposed system exhibited classification accuracy of 88% for atopic dermatitis, 85% for acne vulgaris, and 84.7% for scabies. Again, the computational time (0.0001 seconds) of the proposed system implies that any dermatologist, who decides to implement this study, can attend to not less than 1,440 patients a day compared to the manual diagnosis process. It is estimated that the proposed system will enhance accuracy and offer fasting diagnosis results than the traditional method, which makes this system a trustworthy and resourceful for dermatological disease detection. Additionally, the system can serve as a realtime learning platform for students studying dermatology in medical schools in Ghana.