renardeinside

import numpy as np
import pandas as pd

ds_size = 10000
targets = np.random.choice([0,1],size=ds_size)
probs = np.random.uniform(0,1,size=ds_size)
df = pd.DataFrame([targets,probs]).T
df.columns=['target','proba']

def getLiftTable(df,targetColumn,probaColumn,quantilesSize=10):

renardeinside

import pandas as pd
import sklearn.ensemble as ske
from sklearn.preprocessing import Imputer
from sklearn.feature_selection import SelectFromModel
from sklearn.ensemble import ExtraTreesClassifier,GradientBoostingClassifier
from sklearn.base import TransformerMixin
from sklearn.cross_validation import train_test_split
import numpy as np
from tqdm import tqdm
import xgboost as xgb

renardeinside

renardeinside

# PSI (Population Stability Index) is useful when you want to compare two distributions more gently, then 2-KSA etс
# More docs at http://ucanalytics.com/blogs/population-stability-index-psi-banking-case-study/

import numpy as np
import pandas as pd

def psi(v1,v2,groups=10):
    """
    v1 - first distribution (1-D array)
    v2 - second distribution (1-D array)

renardeinside

renardeinside

import logging

def create_logger(name, log_file, level=logging.INFO,filemode='w'):
    """
    Creates logger
    :param name: logger name
    :param log_file: logger file
    :param level: logging level
    :parameter filemode: output type ('w'-oWerwrite,'a'-Append)

renardeinside

How to setup pip to use local repo?

This method is tested on Linux distribution and based on this sources:

• https://pip.pypa.io/en/stable/user_guide/#config-file (official pip docs)
• https://medium.com/@vando/creating-an-small-local-yum-and-pip-repository-f04ff08c144f (post about)

Steps to setup:

1. Create somewhere on OS file with name pip.conf. This file may also exist in your $HOME/pip.
2. Add this lines to file:

[global]

renardeinside

How to define file encoding on nix system?

Use command:

file -i somefile.csv

In output:

charset=<some charset>

renardeinside

class DataProvider:
  # above goes some low-level code
  def _get_data(self, query: str) -> pd.DataFrame:
      self.logger.debug(f"Running SQL query: {query}")
      start_time = dt.datetime.now()
      data = pd.read_sql(query, self.connection)
      end_time = dt.datetime.now()
      time_delta = end_time - start_time
      self.logger.debug(
          f"Query executed, returning the result. Total query time: {time_delta}"

renardeinside

st.write(
    """
    # Databricks Streamlit Demo :fire:
    This Streamlit application connects to Databricks SQL Endpoint and creates some visualizations based on the [NYC Taxi Dataset](#https://www1.nyc.gov/site/tlc/about/tlc-trip-record-data.page).
    """
)