We should first recall how plasma (intravascular) fluids moves to the extravascular space. It can be approximated by Starling equation. Briefly:
The rate at which fluid is filtered across vascular endothelium (transendothelial filtration) is determined by the sum of two outward forces: capillary pressure ($P_c$) and interstitial protein osmotic pressure ($\pi _{i}$), and two absorptive forces: plasma protein osmotic pressure ($\pi _{p}$) and interstitial pressure ($P_i$).
See more: Starling equation
Primary edema localisation depends on the underlying pathophysiological mechanism that disturb the balance between aforementioned forces pushing fluid outside the vessels. Whether the Starling forces are alerted globally or locally, how long the edema develops etc.
Example from Harisson's Principles of Internal Medicine (I recommend the whole section regarding the edema):
Edema due to an increase in capillary pressure may result from an elevation of venous pressure caused by obstruction to venous and/ or lymphatic drainage. An increase in capillary pressure may be generalized, as occurs in heart failure, or it may be localized to one extremity when venous pressure is elevated due to unilateral thrombophlebitis.
In Congestive Heart Failure edema usually appears in places, where intravascular hydrostatic pressure is the highest. When the individual walks it is typically seen as ankle swelling, when the individual lies down, it is lumbar region edema.
While there is a decrease in osmotic pressure (like in nephrotic syndrome), the fluid moves toward extracellular tissues of relatively "loose" structure (so for example periorbital region).
What is important, in some medical condition (eg. renal failure) different mechanisms may coexists and change the clinical picture.