After some fighting with libpredict, I found that I am going to need a microcontroller with better floating-point support. The Atmega series of microcontrollers do not have an FPU which means any floating-point support is hard (either inaccurate, slow, or hacked).
I found that some of the ARM Cortex processors do indeed have an FPU and their also come in similar form as Arduinos or other development boards:
https://www.adafruit.com/product/2772
Or
https://www.adafruit.com/product/2756
Above is Adafruit’s Feather which I’m going to give a shot. It hosts a Cortex M0 which actually does not have an FPU but would handle floating point much better than Atmega due to 32-bit arch.
The Teensy 3.2 hosts a Cortex M4 which actually does have an FPU which might be the winner.
I’ll be experimenting with both to see which one better suits my needs.