polar 0.0.125

Data Science Library

polar

polar is a Python module that contains simple to use data science functions. It is built on top of SciPy, scikit-learn, seaborn and pandas.

polar build

python setup.py sdist twine upload dist/*

Installation

If you already have a working installation of numpy and scipy, the easiest way to install parkitny is using pip:

pip install polar seaborn pandas scikit-learn scipy matplotlib numpy nltk -U

Dependencies

polar requires:

• Python (>= 3.5)
• NumPy (>= 1.11.0)
• SciPy (>= 0.17.0)
• Seaborn (>= 0.9.0)
• scikit-learn (>= 0.21.3)
• nltk (>= 3.4.5)
• python-pptx (>= 0.6.18)
• cryptography (> 2.8)
• imblearn

Jupyter Notebook Examples

Here is the link to the jupyter notebook with all the exmples that are described below Polar-Examples

ACA (Automated Cohort Analysis) Example

The ACA creates three heatmaps for each feature in the data set.

• Conversion heatmap - conversion per feature value
• Distribution heatmap - distribution per feature value
• Size heatmap - total samples per feature value

Data File: ACA_date.csv

Final Result Power Point: ACA.pptx

import pandas as pd
import polar as pl
from pptx import Presentation
%matplotlib inline

url = "https://raw.githubusercontent.com/pparkitn/imagehost/master/ACA_date.csv"
data_df=pd.read_csv(url)

prs = Presentation()    
pl.create_title(prs,'ACA')
for chart in pl.ACA_create_graphs(data_df,'date','label'):
    pl.add_chart_slide(prs,chart[0],chart[1])
pl.save_presentation(prs,filename = 'ACA')

Conversion:

Distribution:

Samples:

EDA Example

import pandas as pd
import openml
import polar as pl

dataset = openml.datasets.get_dataset(31)
X, y, categorical_indicator, attribute_names = \
dataset.get_data(target=dataset.default_target_attribute,dataset_format='dataframe')

openml_df = pd.DataFrame(X)
openml_df['target'] = y

data_df = pl.analyze_correlation(openml_df,'target')
pl.get_heatmap(data_df,'correlation_heat_map.png',1.1,14,'0.1f',0,100,5,5)

data_df = pl.analyze_association(openml_df,'target',verbose=0)
pl.get_heatmap(data_df,'association_heat_map.png',1.1,12,'0.1f',0,100,10,10)

print(pl.analyze_df(openml_df, 'target',10))

data_df = pl.get_important_features(openml_df,'target')
pl.get_bar(data_df,'bar.png','Importance','Feature_Name')

NLP Example

import nltk
nltk.download('wordnet')
import pandas as pd
import polar as pl
from cryptography.fernet import Fernet

url = "https://raw.githubusercontent.com/pparkitn/imagehost/master/test_real_or_not_from_kaggle.csv"
data_df=pd.read_csv(url)

data_df.drop(columns=['id','keyword','location'], inplace=True)
data_df.head(3)

key = Fernet.generate_key()
data_df['text_encrypted'] =  data_df['text'].apply(pl.encrypt_df,args=(key,))
data_df['text_decrypted'] =  data_df['text_encrypted'].apply(pl.decrypt_df,args=(key,))

data_df['text_stem'] = data_df['text_decrypted'].apply(pl.nlp_text_process,args=('stem',))
data_df['text_stem_lem'] = data_df['text_stem'].apply(pl.nlp_text_process,args=('lem',))

data_df.head(3)

cluster_df = pl.nlp_cluster(data_df, 'text_stem_lem',  10, 'text_cluster',1.0,1,100,1,'KMeans',(1,2))[0]
cluster_df.groupby(['text_cluster']).count()

cluster_df[cluster_df['text_cluster']==9]['text_stem_lem']