I need a large dataset with lattice thermal conductivity , material composition

can any one suggest a dataset with composition,structure,lattice thermal conductivity, or at least composition and thermal conductivity.
is it possible to get lattice thermal conductivity using material.search/summary.search methods?
Thank you

Well, not “large” persay but here’s something: Table of Datasets — matminer 0.8.0 documentation

UCSB thermoelectrics dataset and citrine thermal conductivity datasets

The UCSB one you can likely cross reference with materials project to get structures.

Thank you,
but i have come across this two data sets
i was not able to generate more features with this dataset,
while featurizing it using ElementProperty, i’m getting NAN values,
in data_source i have tried, pymatgen,citrine etc but still not able to generate new features, could you help
here is the code-
from matminer.featurizers.composition.composite import ElementProperty

ep_feat = ElementProperty(data_source=“material project”, features=[“Electronegativity”, “AtomicWeight”, “AtomicRadius”,“Thermal conductivity”,“Heat capacity”,“Atomic number”],
stats=[“minimum”, “maximum”, “range”])
feat10 = ep_feat.featurize_dataframe(df=cotst, col_id=“composition”, ignore_errors=True)

I don’t think there is any data_source called “material project”? If you are featurizing with ignore_errors=True, this is why you are getting nans. Your data_source is being set equal to a string called “material project” which can’t generate any features. You should use one of the presets e.g., “magpie” using the .from_preset featurizer.

See here for the relevant source code: matminer/composite.py at 33bf112009b67b108f1008b8cc7398061b3e6db2 · hackingmaterials/matminer · GitHub

Also, for debugging, setting ignore_errors to False is very helpful because it will tell you what is going wrong. Once you have worked out all the bugs, then doing it with ignore_errors=True is probably what you want.,