Skip to main content

EazeML makes Task of Machine Learning and Data Science super easy.

Project description

GitHub license downloads

Eazeml ;)

drawing1

eazeml is a Python 3.x based Machine Learning Open Source Library which makes the process of machine learning and Data science Faster, Easy and reduces the difficulty of manual coding.

Installation

Use the package manager pip to install eazeml.

pip install eazeml

Usage

import eazeml as ez
help(ez) ## Prints all function and usage

Note:

Only Few Functions example are shown below, for all methods example and usage please see this example notebooks here.

Some Automated functions

quick_ml()

quick_ml method provides automation to data cleaningn & machine learning process here we just need to provide

  1. Pandas Dataframe containg data.
  2. Target-column name we want to make a prediction for.
  3. flag: which specify prediction type (like 'r' for regression and 'c' for classification).
  4. n: n specifies no of features you wan't to use for prediction. it use RFE (Reccurssive Feature Elimination) for extacting best 'n' specified column and make prediction using them.
  • By default 'n' is set to 0 and will use all quick_ml will use all features for prediction if 'n' not specified

and thats it, function will automatically clean & process data and give you output as a complete summary of processing done, Features used and returns report table and model used for prediction

usage:

ez.quick_ml(dataframe,'target_column','flag')

or

report,model = ez.quick_ml(dataframe,'target_column','flag')

or

report,model = ez.quick_ml(dataframe,'target_column','flag',n)
# performs RFE and selects only 'n' number of column for prediction.

for Google Cobal & Kaggle:

ez.plotly_brower() #enables plotly in colab & kaggle.
report,model = ez.quick_ml(dataframe,'target_column','flag',n)
# performs RFE and selects only 'n' number of column for prediction.

quick_pred()

quick_pred method takes cleaned data and prediction flag('r','c') and does prediction using several different algorithms and gives output score using several metrics for each algorithm in tabular format.

example:

  • regression
drawing2
  • classification
drawing1
code:
X = features
y = target column
## flag('r','c')
ez.quick_pred(X,y,'flag')

or

report = ez.quick_pred(X,y,'flag')
## report stores table for future use

deploy_one()

Method Deploys, Machine Learning Model to Production. In one line of code.

Method Creates Fask App which is ready to be deployed to production..

  • User just need to pass
    • Model - Machine Learning Model
    • Features - Features used for Predictions
    • 'title' (optional) - Title to set on Web App. By default it's "Make Predictions".
  • And Done, Your Deployment ready files will be stored in created folder naned 'deployment-files'.
ez.deploy_one(model,features,'titlle')

kaggle()

Method makes participating in kaggle compitions easy.

kaggle method takes input arguments: traindataset, testdataset,target columname and flag('r','c') and gives output as complete summary of steps performed in prediction and data cleaning process and returns predicted values of test dataframe which can be submitted on kaggle.

ez.kaggle(train,test,'targetcolumn','flag')

or

test_pred,report = ez.kaggle(train,test,'targetcolumn','flag')
## test_pred- consist test file prediction
## report - tabular report score

gen_txt_report()

Method genrates txt file named output.txt containig entire summary report of oprations performed

usage:
%%capture example ## captures output of ell and store in example
quick_ml(df,'target','c')
gen_report(example) ## this creates summary file named output.txt
# file contains all log and output of cell

clean_data()

clean_data automatically cleans data by:

  1. removing columns having null value greated than 20%
  2. impute median in missing numeric columns
  3. impute mode in missing categorical column
dataframe = ez.clean_data(dataframe)
## returns cleaned dataframe

nlp_text()

nlp_text converts Textual data of Column to Tfidf vector after operations:

  1. Removing Punctutions
  2. Removing Special Symbol/character
  3. Lemmatization
  4. Coverts Uppercase to lowercase
  • only column passed is converted to vector not whole dataframe
  • need to manually join output data vector with original dataframe
vec = ez.nlp_text(dataframe,'columname')
## returns cleaned Tfidf vector

Some example of Basic functions

importdata()

Method capable of importing tsc,csv,excel file in pandas dataframe object

dataframe = ez.importdata('filename.extension')
## returns cleaned Tfidf vector

info()

info method gives complete information about the data in tabluar format including:

  1. Number of Rows,Column
  2. Number of Categorical, Numerical Feature
  3. null, unique values in each column
  4. total missing value and missing value percentage in each column.
  5. Plots graph of Missing data in percentage format.
ez.info(dataframe)

getcolumnstype()

getcolumnstype returns the type of columns and store it in list format. Saves wasting time in manully writing code and determining columns type.

num,cat = getcolumnstype(dataframe)
## num - list of numeric columns
## cat - list of categorical columns

drop_columns() & dropcolumns()

drop_columns takes argument as Percentage in NUm (int) and columns name

  1. Num - number of null value percentage, above which all columns will be dropped
  2. (optional) columns name - specify columns sepreated with ',' will be dropped

usage:

dataframe = ez.drop_columns(dataframe,n,'columnnames'..)
example:
  • This Example will drop all columns having null value percentage above 20 and drop columns id & name from dataframe.
df = ez.drop_columns(df,n,'id','name')

dropcolumns():

Method just drops specified columns. example:

df = ez.drop_columns(df,'id','name','tp','dsd')

fillnulls()

fillnulls method performs imputation operation of mean ,median,mode or specified value on specified columns together and returns imputed dataframe:

  • Example 1: filling median in columns
df = fillnulls(df,'median','col1','col2',...)
# fill median of respective column to respective column
  • Example 2: filling mean in columns
df = fillnulls(df,'mean','col1','col2',...)
# fill mean of respective column to respective column
  • Example 3: filling mode in columns
df = fillnulls(df,'mode','col1','col2',...)
# fill mode of respective column to respective column
  • Example 4: filling specified value in columns
df = fillnulls(df,'unknown','col1','col2',...)
# unknown will be filled inplace of null values in col2, col1

extract_number()

Method Extracts number from column removing any extra string,special char,symbol, etc and returns cleaned column

df = ez.extract_number(df,'income')
# removes $ symbol and return cleaned column in dataframe

corr_heatmap()

Plots Corelation Heatmap of given dataframe.

arguments:

  1. Dataframe
  2. Style:
    • 'interactive': uses plotly and plots interactive heatmap.
    • 'basic': prints colorized tabular format heatmap.
    • none: uses seaborn to plot annoted heatmap.
ez.corr_heatmap(df)
# Plots heatmap with seaboron
ez.corr_heatmap(df,'interactive')
# plots interactive heatmap with plotly
ez.corr_heatmap(df,'basic')
# prints colorized tabluar correaltion heatmap

confusion_mat()

Method Plots Visualized Confusion Matrix.

examples:

binary

multiclass

#y_true- Actual Value
#y_pred- Predicted Value
ez.confusion_mat(y_true,y_pred)
# you can change color using cmap

roc_curve_graph

Method Plots AUC-ROC plot for y_true,y_pred

auc plot

  • only works with Binary Classification
ez.roc_curve_graph(y_true,y_pred)

box_hist_plot()

method plots interactive Box plot and histogram using plotly library

ez.box_hist_plot(dataframe,'col1','col2',...)
# plots box plot and histogram for all specified column of dataframe.

classification() & regression()

  • classification():

    It is menu based classification, asks user to which model to use for prediction.

    It returns y_predicted,y_actual, model used for classification

    It also prints all classification score metrics and confusion matrix by default

# X = features
# y = target column
y_pred,y_actual,model = ez.classification(X,y)
  • regression():

    It is menu based regression method, asks user for which model to use for prediction.

    It retruns y_predicted,y_actual, model used for prediction

    It also prints all regression score metrics by default

# X = features
# y = target column
y_pred,y_actual,model = ez.regresssion(X,y)

RFE()

Methods Perfroms Recurrsive Feature Elimination and returns best n columns specified

rfecolumns = ez.rfe(model,X_train,y_train,n)
# n- number of features to be selected (defualt n=10)
# model - model to use RFE with.

classification_result() & regression_result()

Prints all scoring Metrics for when passed y_true & y_predicted

  • classification_result()

    Evaluation Matrics:

    • F1 score
    • AUC-ROC score
    • classification report
    • plots confusion matrix
    • plots auc graph (only if binary classification)
ez.classification_result(y_true,y_pred)
  • regression_result()

    Evaluation Matrics:

    • R2 score
    • Mean Squared Error score
    • Root Mean Squared Error score
    • Root Mean Squared log Error score
ez.regression_result(y_true,y_pred)

Machine Learning ALgorithms

Eazeml provides use of each algoritm with custom parameter usage:

# first need to define these variables
X = Feature
y = Targetcolumn
X_train,X_test,y_train,y_test = train_test_split(X,y,train_size=10)
ez.regression_result(y_true,y_pred)
  • These Machine Learning Algorithms are already set to best param grid from experienced users.
  • 'reg' defines regression & 'classifier' defines classifier algorithm.
## pass X_train,y_train,X_test,y_test in which ever algorithm you need.
y_pred,model = ez.linear_reg(X_train,y_train,X_test,y_test)
## return y_predicted and model for further usage.
ez.logistic_reg(X_train,y_train,X_test,y_test)
ez.randomforest_classifier(X_train,y_train,X_test,y_test)
ez.randomforest_reg(X_train,y_train,X_test,y_test)
ez.xgb_classifier(X_train,y_train,X_test,y_test)
ez.xgb_reg(X_train,y_train,X_test,y_test)
ez.lgb_classifier(X_train,y_train,X_test,y_test)
ez.lgb_reg(X_train,y_train,X_test,y_test)
ez.Gradient_boost_classifier(X_train,y_train,X_test,y_test)
ez.Gradient_boost_reg(X_train,y_train,X_test,y_test)
ez.adaboost_classifier(X_train,y_train,X_test,y_test)
ez.adaboost_reg(X_train,y_train,X_test,y_test)

get_IQR

Prints IQR (Inter Quartile Range) values table for each column of dataframe

ez.get_IQR(dataframe)

show_outlier & remove_outlier

  • show_outlier()

Method Prints number of outlier in each column of dataframe

ez.show_outlier(dataframe)
  • remove_outlier()

Method removes all outliers from dataframe and returns Normally Distributed Dataframe and prints IQR range table.

newdf = ez.remove_outlier(dataframe)
# stores outlier free dataframe in newdf

stat_models()

Prints Statastical Significance table(OLS table) for each column of dataframe with respect to target column

X = features
y = target column
ez.stats_models(y,X)

VIF()

Method Prints VIF(Variation Inflation Variance)value of each column in Table format for each column of dataframe

ez.VIF(dataframe)

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for eazeml, version 0.0.7
Filename, size File type Python version Upload date Hashes
Filename, size eazeml-0.0.7-py3-none-any.whl (19.7 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size eazeml-0.0.7.tar.gz (23.8 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page