The epidermal tissues constitute the largest and most prevalent organ in the human body and play a vital role to protect internal organs from external damage. However, these tissues are susceptible to damage due to external factors, and if left untreated, can cause extensive harm. To avoid such complications, it is imperative to identify skin diseases accurately and in a timely manner. Conventional diagnostic methods are often time-consuming and costly, making the use of artificial intelligence essential in addressing this issue. This study proposes a hybrid learning approach for automating skin disease prediction. The approach involves two pre-processing methods: image scaling and image normalization, followed by image classification. Two classification processes were employed, including two ensemble models, gradient boost, and extreme gradient boost in combination with two deep neural models. The best model from the previous two processes was selected and combined to improve the performance of the prediction model. The results indicate that the proposed model’s performance is significantly higher than that of the previous single learning models, achieving an accuracy rate of more than 87%.