What are the properties of building a dateset?

If the data is not evenly distributed between different categories, then which bias will occur?

Is Accuracy the prime factor for classification of fradulant credit card transaction problem?

Data size is more important in Machine Learning than Deep Learning.

In spam detection which metric is the most important factor?

In cancer patient classification which metric is the most important?

In Facial Recognition problem which metrics are most important?

In classification problem when both true positive and true negative are important, then which metric will be best for measurement of that problem?

If San Francesco parking sign dataset is expanded into London, do the dataset need to be trained by new dataset of London?