Look up a resistance vs. temperature table for your NTC (or any NTC for that matter).Use it to calibrate the Steinhart-Hart equation (it can't get better than that).Plot them both, and see how much they deviate.
As said, compare it to a DS18B20. I would trust that more than a calibrating thermistor.An even 16 steps/degreeC all the way, and they also come in a waterproof/encapsulated version.Calibrating that to <0.1 with melting ice (from distilled water), if needed, should be easy.Just add/subtract the difference, if any, before printing.Leo..
Yes.A semicinductor-based sensor with integrated 12-bit A/D (don't tell them that part) doesn't need a precision resistor or a Steinhart-Hart equation that can influence the end result. You just request the temp in degreeC (or degreesF) from the sensor directly.Leo..