Table 6. Impact of Training Set on Linear Regression Parameters for the ¹³C and ¹⁵N Isotropic Shieldings^a.

training set	slope	intercept	RMSE	slope	intercept	RMSE
	13C: Fragment PBE0			15N: Fragment PBE0
1. Benchmark Set	–0.9661	179.45	1.74	–1.0201	197.84	4.93
2. Polymorph Set	–0.9633	178.50	1.59	–0.8987	182.86	0.73
3. Combined Set	–0.9648	179.01	1.63	–1.0209	197.00	4.05
	13C: GIPAW PBE			15N: GIPAW PBE
1. Benchmark Set	–0.9893	169.05	2.46	–1.0165	184.98	9.19
2. Polymorph Set	–0.9663	167.62	2.02	–0.8496	164.87	2.38
3. Combined Set	–0.9746	168.13	2.08	–1.0126	183.30	7.84

^aReported slopes and intercepts were fitted using (1) only the previously published benchmark test set shifts (excluding form I acetaminophen from the ¹³C test set), (2) only the polymorph shifts from this work, or (3) the combination of both the benchmark and polymorph shifts. The RMS errors are reported for the polymorphic crystal ¹³C and ¹⁵N isotropic shifts in the polymorphic crystals studied in this work.