Now we're getting somewhere. If you're getting 50 at ambient (20 C?) then that's 50/1023*5 = 0.244 volts. The voltage divider you are forming has a gain of 1k/(1k+RT) where RT is your thermistor value. The output voltage from your assembly is then 5*1k/(1k+RT)=0.244 volts from which we can solve that RT=20k approximately.
Of course I could have probably suggested you measure that directly with a DMM

If that's true, then you don't have to do any interpolation. I'd replace the 1k resistor with a 10k-20k resistor (let's say 10k for now) then you have the equation:
analog voltage = 5*10k/(10k+RT)
from which you can solve for RT. Thermistors generally have well-known temperature-resistance properties that you can look up at various manufacturers. Given that you only need accuracy to 5 degrees though it would probably be easiest to just build a lookup table of analog voltage reading-->temperature. Use a commercial temperature gauge for the moment to build this lookup table (e.g., heat up your sensor and the temperature gauge at the same time and just take readings).
--
The
Quick Shield: breakout all 28 pins to quick-connect terminals