skwrapper 0.1.5

This package helps the Data Scientist to train there model on dataset with different models without copy pasting the code again and again, this package is a Sklearn wrapper which does performs the model training. this saves time of developer and it also helps with detail metrics and for quick scan which model best fits for dataset

skwrapper

This package helps the Data Scientist to train there model on dataset with different models without copy pasting the code again and again, this package is a Sklearn wrapper which does performs the model training. this saves time of developer and it also helps with detail metrics and for quick scan which model best fits for dataset

Features

• Supports regression and classifications models
• Computes common regression and classificatons metrics.
• Optional display of predicted values.
• Easy-to-use unified interface for training and evaluation.

---

Installation

pip install skwrapper

---

Usage Example

## Import class from Library
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

df = pd.read_csv("Social_Network_Ads.csv")

selected_row = df.loc[:, 'Age': 'Purchased']

from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
# first we have to define X and y where X is the variable or feature input and y is the output target basically
x = selected_row[['Age', 'EstimatedSalary']]
y = selected_row['Purchased']

## Split the Data
X_train, x_test, Y_train, y_test = train_test_split( x, y, train_size=0.8, random_state=48 )

X_train.shape, x_test.shape


# doing standardization
scaler = StandardScaler()

#fit the scaler to the train set, it will learn the parameter
scaler.fit(X_train)  ## learn mean and std from the train dataset
X_train_scaled = scaler.transform(X_train)  ## Apply sacling
X_test_scaler = scaler.transform(x_test) ## Apply same scaling on X_test as well

#convert the numpy 2D arry to pd dataframes with column names on it as numpy array dont have column name after scaling
X_train_scaled_df = pd.DataFrame(X_train_scaled, columns=X_train.columns)  
X_test_scaler_df = pd.DataFrame(X_test_scaler, columns=x_test.columns) 

print(X_train_scaled_df.describe())
print(X_train.describe())


# Train and evaluate Models
|from skwrapper import sc, sr
## Single Model Execution for sc(Supervised Classification Models)
sc.perform(
    case=["logistic"],
    xy_train=[X_train_scaled_df, y_train],
    xy_test=[X_test_scaler_df, y_test],

    ## Optional Parameter
    show_pred=True # If True, predicted_values will be printed. If False, only evaluation metrics will be displayed.
)

## Multiple Model Execution for sc(Supervised Classification Models)
result = sc.perform(
    case=[" logistic", "svc", "knc"],
    xy_train=[X_train_scaled_df, y_train],
    xy_test=[X_test_scaler_df, y_test],

    ## Optional Parameter
    show_pred=True # If True, predicted_values will be printed. If False, only evaluation metrics will be displayed.
)

#----------------and for Supervised Regression Models-------------------#
## Single Model Execution:
sr.perform(
    case=["linearR"],
    xy_train=[X_train, y_train],
    xy_test=[X_test, y_test],

    #Optional Parameter
    show_pred=True # If True, predicted_values will be printed. If False, only evaluation metrics will be displayed.
)

## Multiple Model Execution:
sr.perform(
    case=["linearR", "svr", "knr"],
    xy_train=[X_train, y_train],
    xy_test=[X_test, y_test]
)

print(result) ## This will print all the metrics for all defiend models in **case**

# Access specific model metrics or predection value model object,
print("MSE for Linear Regression:", sc.metrics["linearR"]["mse"])
print("predicted_value for SVR:", sc.metrics["svr"]["predicted_value"])

##############
# dont do this
result = sc.perform( case=["logistic"], xy_train=[X_train_scaled_df, y_train], xy_test=[X_test_scaler_df, y_test],show_pred=True )
print("predicted_value for SVR:", result.metrics["svr"]["predicted_value"]) # result.metrics ❌
##############

## You Can Plot the predicted Values
sns.scatterplot(sc.metrics['svr']['predicted_value'])
plt.show()

---

Using Scikit-learn Parameters

This library acts as a wrapper around scikit-learn models, so you can pass model parameters exactly the same way you would in scikit-learn. All keyword arguments (**kwargs) are forwarded to the underlying scikit-learn model.

Example

from skwrapper import sc, sr
model = sr

model.perform(
    case=["rfr"],
    xy_train=(X_train, y_train),
    xy_test=(X_test, y_test),
    n_estimators=200,
    max_depth=10,
    random_state=42
)

- **Note: Model parameters must match the parameters of the corresponding scikit-learn estimator. Invalid parameters will raise an error.**

Supported Models

• Wrapper class for multiple sklearn regression models.:

Model	Description
linearR	Linear Regression
ridge	Ridge Regression
lasso	Lasso Regression
svr	Support Vector Regression
knr	K-Nearest Neighbors Regressor
gbr	Gradient Boosting Regressor
rfr	Random Forest Regressor
dtr	Decision Tree Regressor

Metrics

• metrics computed automatically:

- Mean Squared Error (MSE)
- Mean Absolute Error (MAE)
- Root Mean Squared Error (RMSE)
- R² Score

---

• Wrapper class for multiple sklearn Classifications models.:

Model	Description
logistic	Logistic Regression Classifier
svc	Support Vector Classifier (SVC)
rfc	Random Forest Classifier
gbc	Gradient Boosting Classifier
knc	K-Nearest Neighbors Classifier
dtc	Decision Tree Classifier

Metrics

• metrics computed automatically:

- accuracy
- confusion_matrix
- classification_report

---