import openml
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import warnings
warnings.filterwarnings('ignore')

# BEGINNER EXAMPLE 1: Load a real customer satisfaction survey dataset
dataset = openml.datasets.get_dataset(42178)
df, _, _, _ = dataset.get_data(dataset_format='dataframe')
print(df.shape)
print(df.head(3))

# BEGINNER EXAMPLE 2: Check for missing values in survey responses
missing = df.isnull().sum()
print(missing[missing > 0])

# BEGINNER EXAMPLE 3: Summarize basic demographic information
print(df['gender'].value_counts())
print(df['SeniorCitizen'].value_counts())

# BEGINNER EXAMPLE 4: Visualize satisfaction scores (churn as proxy)
sns.countplot(data=df, x='Churn')
plt.title('Customer Churn as Satisfaction Indicator')
plt.xlabel('Customer Churned?')
plt.ylabel('Count')
plt.show()

# BEGINNER EXAMPLE 5: Calculate and print churn rate
churn_rate = df['Churn'].value_counts(normalize=True)['Yes'] * 100
print(f'Overall churn rate: {churn_rate:.2f}%')

# BEGINNER EXAMPLE 6: Load a synthetic Net Promoter Score (NPS) survey dataset
np.random.seed(42)
nps_df = pd.DataFrame({'CustomerID': range(1,501),
                       'Age': np.random.randint(18,70,500),
                       'Region': np.random.choice(['North','South','East','West'],500),
                       'NPS_Score': np.random.randint(0,11,500)})
print(nps_df.head(3))

# INTERMEDIATE EXAMPLE 1: Visualize the NPS score distribution
sns.histplot(nps_df['NPS_Score'], bins=11, kde=False)
plt.title('NPS Score Distribution')
plt.xlabel('NPS Score (0-10)')
plt.ylabel('Number of Responses')
plt.show()

# INTERMEDIATE EXAMPLE 2: Calculate NPS according to standard formula
total = len(nps_df)
promoters = len(nps_df[nps_df['NPS_Score'] >= 9])
detractors = len(nps_df[nps_df['NPS_Score'] <= 6])
nps = ((promoters - detractors) / total) * 100
print(f'Net Promoter Score (NPS): {nps:.1f}')

# INTERMEDIATE EXAMPLE 3: Segment NPS by region
region_nps = nps_df.groupby('Region').apply(
    lambda d: ((len(d[d['NPS_Score'] >= 9]) - len(d[d['NPS_Score'] <= 6])) / len(d)) * 100
)
print(region_nps)

# INTERMEDIATE EXAMPLE 4: Visualize regional NPS differences
region_nps.plot(kind='bar', color='skyblue')
plt.title('Regional Net Promoter Score (NPS)')
plt.xlabel('Region')
plt.ylabel('NPS')
plt.ylim(-100, 100)
plt.show()

# INTERMEDIATE EXAMPLE 5: Load and preview open-ended customer feedback data
feedback_df = pd.DataFrame({
    'CustomerID':[1,2,3,4,5],
    'Feedback':['Great service and friendly staff',
                'Delivery was slow and packaging was poor',
                'Excellent quality, will buy again',
                'Customer support needs improvement',
                'Good value for money']
})
print(feedback_df.head())

# INTERMEDIATE EXAMPLE 6: Simple sentiment identification
feedback_df['Sentiment'] = feedback_df['Feedback'].str.contains('great|excellent|good|friendly|buy again', case=False).map({True: 'Positive', False: 'Negative'})
print(feedback_df[['Feedback', 'Sentiment']])

# ADVANCED EXAMPLE 1: Cross-tabulation between Churn and Internet Service
ct = pd.crosstab(df['InternetService'], df['Churn'], normalize='index') * 100
print(ct)

# ADVANCED EXAMPLE 2: Monthly churn trend analysis
df['SignupMonth'] = (df['tenure'] // 1).astype(int)
monthly_churn = df.groupby('SignupMonth')['Churn'].apply(lambda x: (x=='Yes').mean() * 100)
monthly_churn.plot(kind='line', marker='o')
plt.title('Monthly Churn Rate Trend')
plt.xlabel('Month Since Signup')
plt.ylabel('Churn Rate (%)')
plt.show()

# ADVANCED EXAMPLE 3: Create a Customer Satisfaction Index
satisfaction_index = df[['MonthlyCharges', 'tenure']].apply(lambda x: (x['MonthlyCharges']/x['tenure']) if x['tenure'] else x['MonthlyCharges'], axis=1)
df['SatisfactionIndex'] = satisfaction_index
print(df[['MonthlyCharges', 'tenure', 'SatisfactionIndex']].head(3))

# Median Imputation for Numeric Columns
df_filled = df.copy()
num_cols = ['TotalCharges', 'MonthlyCharges']
df_filled[num_cols] = df_filled[num_cols].apply(
    pd.to_numeric, errors='coerce'
)
df_filled[num_cols] = df_filled[num_cols].fillna(
    df_filled[num_cols].median()
)
print(df_filled[num_cols].isnull().sum())

# ERROR HANDLING 2: Guard against grouping errors
try:
    result = df.groupby('Gender').size()
except KeyError:
    print('Check column spelling: Should be gender (lowercase), not Gender.')

# ERROR HANDLING 3: Watch out for misinterpreted NPS scale
def recode_nps(score):
    try:
        s = int(score)
        if s > 10 or s < 0:
            return np.nan
        else:
            return s
    except:
        return np.nan
nps_df['NPS_Score'] = nps_df['NPS_Score'].apply(recode_nps)
print(nps_df['NPS_Score'].isnull().sum())

# PATTERN EXAMPLE 1: Customer segmentation by contract type
seg_ct = pd.crosstab(df['Contract'], df['Churn'], normalize='index') * 100
print(seg_ct)

# PATTERN EXAMPLE 2: Index construction for advanced benchmarking
df['ChargeIndex'] = df['MonthlyCharges']/df['MonthlyCharges'].max()
df['TenureIndex'] = df['tenure']/df['tenure'].max()
print(df[['ChargeIndex', 'TenureIndex']].head(3))

# FULL MARKET RESEARCH PROBLEM: From survey data to a clear recommendation
dataset = openml.datasets.get_dataset(42178)
df, _, _, _ = dataset.get_data(dataset_format='dataframe')
df['MonthlyCharges'] = df['MonthlyCharges'].fillna(df['MonthlyCharges'].median())
nps_proxy = (df['Churn'].map({'No': 10, 'Yes': 0}))
high_value = df['MonthlyCharges'] > df['MonthlyCharges'].median()
promoters_pct = (nps_proxy[high_value] >= 9).sum() / high_value.sum() * 100
print(f'Percentage of high-value customers who are promoters: {promoters_pct:.2f}%')