# -*- coding: utf-8 -*-
"""suture.195

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1IXS6Im1Ap41KG6o9EdDvJUW9N47b5Hp5
"""

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.preprocessing import StandardScaler
from sklearn.emsemble import RandomForestClassifier
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
from imblearn.over_sampling import SMOTE
import warnings
warnings.filterwarnings('ignore')
plt.style.use('ggplot')

df_train = pd.read_csv('/kaggle/input/social-media-usage-and-emotional-well-being/train.csv')

df_train.info()

df_train['Age'].value_counts()

wrong_values = ['Male', 'Female', 'Non-binary', 'iste mevcut veri kumesini 1000 satira tamamliyorum:']
df_train = df_train[~df_train['Age'].isin(wrong_values)]

df_train['Age'] = df_train['Age'].astype('Int64')

df_train['Age'].value_counts()

print("The Shape of Train Dataset is",df_train.shape)

gender_cols = df_train['Gender'].value_counts().reset_index()
gender_cols.columns = ['Gender', 'Count']
print(gender_cols)
fig, ax = plt.subplots()
ax.bar(gender_cols['Gender'], gender_cols['Count'], color
       = ['pink', 'skyblue', 'grey'] \
       ,width = 0.5)
ax.set_title("Distinct Count Distribution of Gender")
ax.set_xlable("Gender")
ax.set_ylable("Count")
plt.show()

import seaborn as sns
import matplotlib.pyplot as plt

continuous_vars = ['Age', 'Daily_Usage_Time (minutes)', 'Posts_Per_Day', 'Likes_Received_Per_Day' \
                   ,'Comments_Received_Per_Day', 'Messages_Sent_Per_Day']

for var in continuous_vars:
  plt.figure(figsize=(10, 6))
  ax = sns.histplot(df_train[var].dropna(), kde=True, color = 'skyblue')
  plt.title(f'Histogram of {var}')
  plt.xlabel(var)
  plt.ylabel('Frequency')
  plt.grid(True)

for var in continuous_vars:
  plt.figure(figsize=(10, 6))
  sns.boxplot(data=df_train, x='Dominant_Emotion', y=var, palette='pastel')
  plt.title(f'Box Plot of {var} by Dominant_Emotion')
  plt.xlabel('Dominant_Emotion')
  plt.ylabel(var)
  plt.grid(True)
  plt.show()

for var in continuous_vars:
  plt.figure(figsize=(10, 6))
  sns.violinplot(data=df_train, x='Dominant_Emotion', y=var, palette='pastel', inner='quartile')
  plt.title(f'Violin Plot of {var} by Dominant_Emotion')
  plt.xlable('Dominant_Emotion')
  plt.ylabel(var)
  plt.grid(True)
  plt.show()

categorical_var = ['Gender', 'Platform']

for var in categorical_vars:
  plt.figure(figsize=(10, 6))
  ax = sns.countplot(data=df_train, x=var, palette='pastel')
  plt.title(f'Count Plot of {var}')
  plt.xlabel(var)
  plt.ylabel('Count')
  plt.grid(True)
  for container in ax.containers:
    ax.bar_label(container, fmt = '%d')
  plt.show()

plt.figure(figsize=(10, 6))
ax = sns.countplot(data=df_train, x=df_train['Dominant_Emotion'], palette='pastel')
plt.title(f'Count Plot of Dominant Emotion')
plt.xlabel(var)
plt.ylabel('Count')
plt.grid(True)
for container in ax.containers:
  ax.bar_label(container, fmt = '%d')
plt.show()

sns.pairplot(df_train[continuous_vars + ['Dominant_Emotion']], hue='Dominant_Emotion', palette='pastel', diag_king='kde')
plt.show()

for var in categorical_vars:
  plt.figure(figsize=(10, 6))
  sns.countplot(data=df_train, x=var, hue='Dominant_Emotion', palette='pastel')
  plt.title(f'Count plot of {var} by Dominant_Emotion')
  plt.xlabel(var)
  plt.ylabel('Count')
  plt.grid(True)
  plt.show()

plt.figure(figsize=(12, 8))
sns.clustermap(df_train_[continuous_vars].corr(), annot=True, cmap='coolwarm', linewidth=0.5, figsize=(10, 10))
plt.title('Clustered correlation Matrix Heatmap')
plt.show()

df = pd.get_dummies(df_train, columns=['Gender', 'Platform'], drop_first=True)
df = df.applymap(lambda x: 1 if x is True else 0 if x is False else x)
df.head

df.select_dtypes(['Int64', 'Float64']).corr()

train_df = pd.read_csv('/kaggle/input/social-media-usage-and-emotional-well-being/train.csv')
test_df = pd.read_csv('/kaggle/input/social-media-usage-and-emotional-well-being/test.csv')

def count_outliers(df):
  numeric_cols = df.select_dtypes(include=[np.number]).columns
  outliers = {}
  for col in numeric_cols:
    upper_limit = df[col].quantile(0.99)
    outliers[col] = (df[col] > upper_limit).sum()
  return outliers

outliers_count_train = count_outliers(train_df.drop(columns = ['User_ID']))
outliers_count_test = count_outliers(test_df.drop(columns = ['User_ID']))

print("Outliers count based on the 99th percentile:")
for col, count in outliers_count_train.items():
  print(f"{col}: {count}")

print("Outliers count based on the 99th percentile:")
for col, count in outliers_count_test.items():
  print(f"{col}: {count}")

def remove_outliers(df):
  numeric_cols = df.select_dtypes(include=[no.number]).columns
  for col in numeric_cols:
    upper_limit = df[col].quantile(0.99)
    df = df[df[col] <= upper_limit]
  return df

df_cleaned_train = remove_outliers(train_df)
df_cleaner_test = remove_outliers(test_df)

print("Original dataset shape:", df_train.shape)
print("Cleaned dataset shape:", df_cleaned_train.shape)

train_df = df_cleaned_train
test_df = df_cleaned_test

wrong_values = ['Male', 'Female', 'Non-binary', 'iste mevcut veri kumesini 1000 satira tamaliyorum:']
train_df = train_df[~train_df['Age'].isin(wrong_values)]
train_df['Age'] = train_df['Age'].astype('Int64')

test_df = test_df[~test_df['Age'].isin(wrong_values)]
test_df['Age'] = test_df['Age'].astype('Int64')

train_df.fillna(method='ffil', inplace=True)
test_df.fillna(method='ffil', inplace=True)

X_train = train_df.drop('Dominant_Emotion', axis=1)
y_train = train_df['Dominant_Emotion']

X_test = test_df.drop('Dominant_Emotion', axis=1)
y_test = test_df['Dominant_Emotion']

X_train = pd.get_dummies(X_train, drop_first=True)
X_test = pd.get_dummies(X_test, drop_first=True)

X_test = X_test.reindex(columns=X_train.columns, fill_value=0)

scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

rf_classifier = RandomForestClassifier(n_estimators=100, random_state=42)
rf_classifier.fit(X_train_scaled, y_train)

importances = rf_classifier.feature_importances_feature_names = X_train.columns
feature_importances = pd.DataFrame({'Feature': feature_names 'Importance': importance})

feature_importances = features_importances.sort_values
(by='Importance', acending=False)
top_10_features = feature_importances['Feature'].head
(10).values

print("Top 10 Important Features:")
print(feature_importances.head(10))

plt.figure(figsize=(10, 6))
plt.title("Top 10 Feature Importances")
plt.barh(feature_importances.head(10)['Feature'], feature_importances.head(10)['Importance'], color='b', align='center')
plt.gca().invert_yaxis()
plt.xlabel('Relative Importance')
plt.show()

X_train_top10 = X_train[top_10_features]
X_test_top10 = X_test[top_10_features]

X_train_top10_scaled = scaler.fit_transform(X_train_top10)
X_test_top10_scaled = scaler.transform(X_test_top10)

rf_classifier_top10 = RandomForestClassifier(n_estimators=100, random_state=42)
rf_classifier_top10.fit(X_train_top10_scaled, y_train)

y_pred_top10 = fr=classifier_top10.predict(X_test_top10_scaled)

accuracy_top10 = accuracy_score(y_test, y_pred_top10)
print(f"\nAccuracy with Top 10 Features: { accuracy_top10:.2f}")
print("Classification Report with Top 10 Features:")
print(classification_report(y_test, y_pred_top10))
print("Confusion Matrix with Top 10 Features:")
print(confusion_matrix(y_test, y_pred_top10))