RandomForestClassifier.py

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from warnings import filterwarnings
filterwarnings('ignore')
pd.set_option('display.max_columns', None)
pd.set_option('display.max_colwidth', None)
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.svm import SVC
from sklearn.metrics import mean_absolute_error,r2_score,accuracy_score,classification_report,confusion_matrix
import pickle

df=sns.load_dataset('tips')
print(df.head())
# we will predict time based on other info
print(df.time.unique())
# Feature Engineering: Handling Missing values, categorical features, outliers, feature scaling, Automate this process
print(df.info())
from sklearn.preprocessing import LabelEncoder
encoder=LabelEncoder()
df['time']=encoder.fit_transform(df[['time']])
print(df.head())
print(df.time.value_counts())
# Independent and dependent features
X=df.drop(labels=['time'],axis=1)
y=df['time']
print(X.head())
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.2,random_state=42)
from sklearn.impute import SimpleImputer
from sklearn.preprocessing import OneHotEncoder
from sklearn.preprocessing import StandardScaler
from sklearn.pipeline import Pipeline
from sklearn.compose import ColumnTransformer
categorical_cols=['sex', 'smoker', 'day']
numerical_cols=['total_bill','tip','size']
# Feature ENgineering Pipeline
num_pipeline=Pipeline(
    steps=[
        ('imputer',SimpleImputer(strategy='median')),#missing value
        ('scaler',StandardScaler())#feature scaling
    ]
)
cat_pipeline=Pipeline(
    steps=[
        ('imputer',SimpleImputer(strategy='most_frequent')),#missing value
        ('onehotencoder',OneHotEncoder())#categorical feature to numerical
    ]
)
preprocessor=ColumnTransformer([
    ('num_pipeline',num_pipeline,numerical_cols),
    ('cat_pipeline',cat_pipeline,categorical_cols)
])
X_train=preprocessor.fit_transform(X_train)
X_test=preprocessor.transform(X_test)
from sklearn.ensemble import RandomForestClassifier
# Automate Model training process
models={
    'Random Forest':RandomForestClassifier(),
    'Decision Tree': DecisionTreeClassifier(),
    'SVC': SVC()
}
from sklearn.metrics import accuracy_score
def evaluate_model(X_train,y_train,X_test,y_test,models):
    report={}
    for i in range(len(models)):
        model=list(models.values())[i]
        # Train model
        model.fit(X_train,y_train)
        y_pred=model.predict(X_test)
        test_model_score=accuracy_score(y_test,y_pred)
        report[list(models.keys())[i]]=test_model_score
    return report
print(evaluate_model(X_train,y_train,X_test,y_test,models))
classifier=RandomForestClassifier()
# Hyperparameter tuning
params={'max_depth':[3,5,10,'None'],
        'n_estimators':[100,200,300],
        'criterion':['gini','entropy']}
from sklearn.model_selection import RandomizedSearchCV
cvf=RandomizedSearchCV(classifier,param_distributions=params,cv=5,scoring='accuracy',verbose=3)
cvf.fit(X_train,y_train)
print(cvf.best_params_)