I have very sparse data in a pandas dataframe with 25million+ records. This has to be converted into a multi dimensional numpy array. I have written this the straightforward way using a for
loop, and was wondering if there is a more efficient way.
import numpy as np
import pandas as pd
facts_pd = pd.DataFrame.from_records(columns=['name','offset','code'],
data=[('John', -928, 'dx_434'), ('Steve',-757,'dx_5859'), ('Jack',-800,'dx_250'),
('John',-919,'dx_401'),('John',-956,'dx_5859')])
name_lu = pd.DataFrame(sorted(facts_pd['name'].unique()), columns=['name'])
name_lu["nameid"] = name_lu.index
offset_lu = pd.DataFrame(sorted(facts_pd['offset'].unique(), reverse=True), columns=['offset'])
offset_lu["offsetid"] = offset_lu.index
code_lu = pd.DataFrame(sorted(facts_pd['code'].unique()), columns=['code'])
code_lu["codeid"] = code_lu.index
facts_pd = pd.merge(pd.merge(pd.merge(facts_pd, name_lu, how="left", on="name")
, offset_lu, how="left", on="offset"), code_lu, how="left", on="code")
facts_pd.drop(["name","offset","code"], inplace=True, axis=1)
facts_np = np.zeros((len(name_lu),len(offset_lu),len(code_lu)))
for row in facts_pd.iterrows():
i,j,k = row[1]
facts_np[i][j][k] = 1
Refurbished code:
import numpy as np
import pandas as pd
facts_pd = pd.DataFrame.from_records(columns=['name','offset','code'],
data=[('John', -928, 'dx_434'), ('Steve',-757,'dx_5859'), ('Jack',-800,'dx_250'),
('John',-919,'dx_401'),('John',-956,'dx_5859')])
facts_np = facts_pd.as_matrix()
print facts_np # Displays the data frame in numpy array format.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.