tab data format

[gpagliuca]

the tab.data dataframe is difficult to handle (properties as index), its structure should be changed

[kjeeetil]

xarray provides a very convenient format for this type of data. if you pass a pandas dataframe with an appropriate multiindexed dataframe as input everything works out nicely:

tab=pyfas.tab.Tab("PVTtable.tab")
tab.export_all()

columns=tab.data["Property"].values
data=tab.data["values"].values

series={}
for i, array in enumerate(data): #cycles through each column in data, creating a pandas dataseries with properties as columns
series[columns[i]]=array

new_p=[]
new_t=[]

for p in tab.metadata['p_array'][0]:
new_t=new_t+list(tab.metadata['t_array'][0])
new_p=new_p+[p/1e5 for x in tab.metadata['p_array'][0]]

index=zip(new_p, new_t) #creates a list of tuples which can be used in a multiindex
index=pd.MultiIndex.from_tuples(index, names=["P", "T"]) #creates a multiindex with pressure and temperature

df=pd.DataFrame(series, index=index) #creates a dataframe from the tab data with pressure and temperature as index
da=df.to_xarray() #passes the dataframe to an xarray dataarray

The dataarray will have pressure and temperature as dimensions, and each column in the dataframe as a variable. The datarray allows you to do things like fast linear interpolation (e.g. da.interp(T=50, P=[50, 60, 70])["LIQUID DENSITY"]) which is extremely useful especially when comparing tables with different spacing between input parameters.

[gpagliuca]

thanks for the suggestion, it looks very interesting. Also here if you have the possibility a PR would be quite useful since you know xarray better than me

[daniel-brett]

Has this issue been addressed with #23?

[gpagliuca]

Yes, it was Cheers Il giorno mer 26 apr 2023 alle ore 15:18 daniel-brett < ***@***.***> ha scritto:

…

Has this issue been addressed with #23 <#23>? — Reply to this email directly, view it on GitHub <#12 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AD2WTVRC6D2XNHTBUU7YNI3XDEOC3ANCNFSM4CQ5NGUA> . You are receiving this because you authored the thread.Message ID: ***@***.***>