Top Highlights

• depending on this threshold, your value of precision and recall can change drastically.
• This type of encoding cannot be used in linear models, support vector machines or neural networks as they expect data to be normalized (or standardized).
• We see that all categories in this column have a significant count.
• The rule is simple. If it’s a standard classification problem, choose stratified k-fold blindly.
• set, we have 90 non- pneumothorax and 10 pneumothorax images. If you say that all images in the validation set are non-pneumothorax, what would your accuracy be? Let’s see; you classified 90% of the images correctly. So, your accuracy is 90%.