Datasets:

cannlytics

/

cannabis_results

like 8

Follow

Cannlytics 12

This pull request incorporates improvements found during a rigorous testing of CoADoc: the parsing of a dataset of public cannabis lab test results published by Raw Garden. There's a whole lot of cleaning to be done on this data! For example, there's a value of '<LLoQ', and I found 200+ more! If anyone else wants to help look for odd values in any columns or datasets, then below are version 1.0.0 of the Raw Garden lab result data.

1. 'rawgarden_details'
2. 'rawgarden_results'
3. 'rawgarden_values'

Example usage:

import pandas as pd

# Read results.
df = pd.read_csv('https://cannlytics.page.link/?link=https://firebasestorage.googleapis.com/v0/b/cannlytics.appspot.com/o/data%252Flab_results%252Frawgarden%252Fresults.csv?alt%3Dmedia%26token%3Ddd868e72-edde-4278-9725-b33368a35d54')

# Look at all of the unique values.
unique_values = list(df['value'].unique())
errors = []
for value in unique_values:
    try:
        pd.to_numeric(value)
    except:
        errors.append(value)
print(errors)

Furthermore, the new utility function, create_hash, found in cannlytics.utils.utils.py, needs to be incorporated into CoADoc's standardize and perhaps save / aggregate methods. For now, users can perform hashing for their data as follows.

from cannlytics.utils.utils import create_hash
import pandas as pd

# Read details.
df = pd.read_csv('https://cannlytics.page.link/?link=https://firebasestorage.googleapis.com/v0/b/cannlytics.appspot.com/o/data%252Flab_results%252Frawgarden%252Fdetails.csv?alt%3Dmedia%26token%3De5b5273a-049a-4092-98d7-90a62ef399a3')

# Create hashes for a given DataFrame of details.
df= df.where(pd.notnull(df), None)
df['results_hash'] = df['results'].apply(lambda x: create_hash(x))
df['sample_hash'] = df.loc[:, df.columns != 'sample_hash'].apply(
    lambda x: create_hash(x.to_dict()),
    axis=1,
)
datafile_hash = create_hash(df)